PowerPoint Presentation


1 The New Physics of Financial Services Understanding how artificial i ntelligence is transforming the financial ecosystem Part of the Future of Financial Services series | Prepared in collaboration with Deloitte August 2018

2 Introductions Foreword multistakeholder Consistent with the World Economic Forum’s mission of applying a approach to address issues of global impact, Forum’s creating this report involved extensive outreach and dialogue with numerous organizations and individuals. They included the from academia and the public sector. The outreach Financial Services, Innovation and Technology communities, and professionals international workshop sessions, encouraging collaborative dialogue to discuss insights and six involved over 200 interviews and rtificial i ntelligence within the financial services industry. opportunities concerning the impact of a content of this report would not be as rich without the support of, and contributions from, the subject matter The holistic and global in experts who assisted our thoughts about the future of the impact of AI on the future of the financial services industry. In shaping , who particular, we thank this project’s Steering Committee and Working Group played an invaluable role with their expertise and patient World Economic Forum and the leadership mentorship. Also critical has been the ongoing institutional support for this initiative from the of our chairman, whose vision of the Fourth Industrial Revolution has been inspirational to this work. 1 network, for its generous Finally, we are grateful to Deloitte Consulting LLP in the United States, an entity within the Deloitte commitment and support in its capacity as the official professional services adviser to the World Economic Forum for this project. Contact 1 private to one or more of Deloitte Touche Tohmatsu Limited (a UK Deloitte refers company limited by guarantee [“DTTL”]), its network of member firms and their related entities. DTTL and each of its For feedback or questions, please to clients. Please see member are legally separate and independent entities. DTTL, also referred to as “Deloitte Global”, does not provide services firms www.deloitte.com/about for a more detailed contact: Deloitte may not be available description of DTTL and its member firms. Please see subsidiaries. its LLP and Certain services of legal structure description of the detailed a for www.deloitte.com/us/about and to public of regulations accounting. rules under the clients attest R. Jesse McWaters, Lead Author financial, business, accounting, rendering publication, this of means by . is and only information general contains publication This Deloitte services or advice professional other or tax legal, investment, not, should it be used as a basis for any decision or action that may affect your business . Before making any decision or taking any This publication is not a substitute for such professional advice or services, nor [email protected] loss publication this on relies who person any by sustained any for responsible be not shall . . adviser professional qualified a consult should you business, your affect may that action Deloitte +1 (646) 623 4500 ( Team Heritage Hubble the and ESA , /AURA NASA : credit image Cover ) STScI 1 The New Physics of Financial Services |

3 Introductions Editors’ note has emerged as a clear focus of discussions at the World Economic Forum’s Annual Meeting Artificial intelligence is a critical aspect of the Fourth Industrial Revolution and over the past few years. large - scale investments in AI, while governments and regulators seek to grapple with the significant Financial institutions around the world are making uncertainties and growing public trepidation as AI becomes central to the fabric of institutions and markets. multistakeholder The Forum has a successful track record of providing detailed World Economic analysis of the changing landscape of the financial ecosystem, particularly Future was clear that a similar approach could cut through the sensationalism surrounding AI to provide valuable insights for the t series. Services through our I of Financial private sector and policy - makers alike. and through this process we have discovered that the long - term Over the past year, we have engaged in what may be the largest study of its kind into AI in financial services, imagined impacts of AI may be even more radical and transformative than we first . Indeed, the central thrust of the document that follows is that the very fabric of the financial , catalysed reorganization in large part by the capabilities and requirements of AI. services ecosystem has entered a period of Our hope is that this document helps you and your institution make informed decisions about how to interpret the evolving role of AI in financial services and navigate the turbulent changes on the horizon. With regards, Rob Galaski R. Jesse McWaters Global Leader, Banking & Capital Markets Project Lead, Future of AI in Financial Services Deloitte Consulting World Economic Forum Past reports from the Future of Financial Services Series (2017) (2016) (2016) (2015) 2 The New Physics of Financial Services |

4 Committee Steering Members of the Steering Committee Nick Cafferillo Olivier Bouée - Pierre Solmaz Altin Chief Digital Chief Operating Officer, Officer, Chief Technology Officer, Credit Suisse S&P Global Allianz SE Vanessa Colella Robert Contri Juan Colombas Head of Citi Ventures & Chief Global Financial Services Leader, Chief Operating Officer, Innovation Officer Lloyds Banking Group Citi , Deloitte Alain Deschênes David Craig Rob Goldstein Senior Vice President and Chief President, Financial & Risk, Chief Operating Officer, Thomson Reuters Operations Officer, PSP Investments BlackRock Prof. Dr Axel P. Lehmann Rakshit Kapoor Ashwin Kumar Group Chief Data Officer, President Personal & Corporate Group Head of Product UBS HSBC Banking, AG Börse Deutsche Development, JP Rangaswami Max Neukirchen Daniel Nadler Head of Strategy, Chase Chief Data Officer; Head of Strategy Officer, Chief Executive JP Morgan Kensho and Innovation, Deutsche Bank Kush Saxena Michael Zerbs Officer, Chief Technology Officer, Chief Technology Scotiabank Mastercard 3 The New Physics of Financial Services |

5 Working Group Members of the Working Group Secil Arslan Beth Devin Tim Baker Head of R&D Special Projects and Head of Innovation Network, Global Head of Innovation, Financial Citi & Risk, Thomson Reuters Yapi Kredi AI / ML, Milos Krstajic Roland Fejfar Gero Gunkel – FinTech IBD, Executive Director Group Head of Artificial Intelligence, Data Scientist at Allianz Group, Morgan Stanley Zurich Insurance Allianz SE - Lin Lee Lena Wei Juan Martinez Phd , Cresnik - Mass Managing Director – America and UK , Strategy and BlackRock Senior Director PayPal Growth, Region, SWIFT Michael Jim Psota O’Rourke Jennifer Peve Co & Chief Technology Founder Head of Machine Intelligence and - Head of FinTech - Co Strategy, Officer, Data Services, NASDAQ Panjiva (S&P Global) DTCC Chadwick Westlake Annika Schröder Nicolas de Skowronski Senior Vice President, Structural Cost of Advisory Excellence, Deputy AI Program Lead, Group Innovation, Head Investment Solution Group, Head, UBS Transformation & Lean, Scotiabank Bank Julius Baer & Co. Ltd 4 The New Physics of Financial Services |

6 Project Team Team Members of the Project Project Authors Project Leadership Future of AI in Financial Services project leadership team includes individuals The following the to its gratitude expresses Economic Forum World The the on the following individuals: team: project Canada Deloitte LLC Forum World Economic Lead McWaters, Jesse R. d Lea Author, Project Courtney Kidd Chubb, Senior Manager Matthew Blake, Head of Financial and Monetary System Initiative Denizhan Uykur, Senior Consultant from Professional Services Deloitte Special thanks for contributions from: Leadership Advis e r Allianz Rob Galaski, Co ‐ Alexandra Blickling , Author, Project Thorsten Münch, Börse Deutsche Additional thanks team expresses gratitude to the following individuals for their contributions and support: The project Emma Barton Jenny Pan Brassard Alexandra Romic Gabriel Kerry Sachdev Siddhant Butts Steven Siegel Alexandra Durbak Maha Eltobgy Elaine Smith Nadia Guillot Hemanth Soni Hart Tiffany Ferdina Yarzada Abel Lee Yik Han 5 The New Physics of Financial Services |

7 Table of contents Preface Context and approach 7 Background The new physics of financial services 17 Executive summary Chapters financial services? How will AI reshape 22 findings Key What are the opportunities and challenges for implementing AI in financial services today? 55 Opportunities, challenges and broader societal i mplications of financial services? How is AI being deployed by sectors 85 explorations Sector How might the continued evolution of AI affect the future of financial services? 142 scenarios for if?” the long Selected “what - term implications of AI for financial services Concluding thoughts 153 Next steps for the financial services ecosystem Additional Reading 155 Acknowledgements 157

8 Context and approach

9 Context and approach | History of AI Unlike past ‘AI Springs’, the science and practice of AI appears poised to continue an unprecedented multi - decade run of advancement 1 ime AI development t ver o “Internet of Things” Semiconductor microprocessors Mainframes Big data Cloud storage and processing 5 More “things or objects” than people AWS) launches John Mashey popularizes the term – Amazon “big Web Services ( 2008 1971 – Intel launches the Intel 4004 chip, known – – 2006 1953 – IBM unveils the IBM 701 “Defense 2009 – 1998 4 3 2 6 connected as the first commercially available microprocessor are now Calculator” mainframe to the internet data” Science fiction AI spring AI winter AI AI spring AI spring winter 2010 1990s 1980s 1970s 1960s – Present 1950s 2000s publishes Turing With funding from sources − Lack of computing − Chess grand master Garry 2011 − – − John Hopfield and David releases Siri, Apple − S tanford robot wins the − − Alan 7 is defeated Computing Machinery and DARPA Grand by Kasparov Challenge the first modern popularize Rumelhart limits the progress power such as DARPA, “virtual “deep learning ” academics lead a boom in IBM’s Deep Blue assistant” Intelligence , which posits a of AI technologies, drying Hinton and others Geoffrey − techniques AI research up funding logical framework on how − − repopularize 2016 – Dragon Systems speech Google to build “intelligent” Deepmind’s recognition software is − Japanese and Simon’s Newell for − government “backpropagation” AlphaGo, machines and test their 8 nets Windows invests $400 million in AI belief deep implemented on General Problem Solver using trained (the Imitation intelligence “reinforcement and related technology Game) − − Joseph Weizenbaum’s – Google Brain uses 2009 learning” , defeats Lee ELIZA GPUs to create capable 10 Go Sedol on a game of 9 deep n eural ets n ow? What’s next on the horizon? What is different n Technology advancement: Enabling technologies have provided both the data and the Current AI technologies lack the ability of “abstraction”, which is the earning: Transfer l • • ability to take lessons from one area and apply them to another access to cheap processing power that was lacking in past AI winters Talent and focus: AI development today is being led by technologists rather than • earning: l Efficient Deep learning shows promise in reducing the effort required to • eep d train models by automating feature extraction. However, heavy data requirements make academics. Beyond just proofs of concept or trials, software engineers are building real apps myriad these techniques impractical for common uses. New developments will increase the for the real successful - use cases across the economy world, leading to efficiency of these algorithms with immediate - term profitability • Leadership and investment: Big technology firms, betting heavily are requirements, • Unstructured learning: on AI, demonstrating the belief that these technologies will in In the pursuit of general intelligence, advancements - simulation reinforcement learning and other based modelling techniques are opening up new improve profitability areas in which AI can be used 8 The New Physics of Financial Services |

10 approach | Focus on AI Context and The public discourse on AI in financial services is highly sensationalized, creating an excess of both exuberance and fear Tremendous excitement is driving today’s m artificial i ntelligence ” oment “ Tepid but significant Significant cross - A top priority for f industry Sustained and strong inancial investment from financial investment s ervice e xecutives investment growth institutions ~$10 ~$58 76% billion billion 48% CAGR Investment Global cross sector growth in that financial in AI by - Global AI i nvestment by of banking CXOs agree 11 12 11 institutions by 2021 AI will be critical to adopting AI investment through 2021 2020 their organization’s ability to 13 market differentiate in the significant uncertainty this excitement is also coupled with However, AI investing could be the next crisis 14 Jobs Robots Could Steal 40% of Sensationalism risks dampening the benefits that AI could bring to financial services, while exacerbating the harms 9 The New Physics of Financial Services |

11 Context and approach | The AI effect Conversations with global experts and business leaders reveal a lack of common effect ’ definitions for AI this is an example of the ‘AI – ntelligence ” is or isn’t No one can agree on what “ artificial i AI, There is a marked lack of clarity around the definition of which frequently leads to confusion and outright disagreement. In our interviews with stakeholders we found variations in experts’ definitions of AI, irrespective of their technical background or formal education in computer science and computer slight – and sometimes not - so - slight – engineering. Selected quotes from i nterviews “What we’re “AI is really talkin g about is about – deep learning “Machine augmented no one has learning i sn’t intelligence” “Machine achieved AI ” AI!” “Cognitive is the i a learning s human aspect of subset of AI” AI” The AI effect This lack of definitional clarity is illustrative of a well - is, and to agree on what of observers inability Essentially, this means the ffect”. e AI “ he t called phenomenon documented isn’t, intelligence and a tendency to conclude that the existing capabilities illustrates this Author Pamela McCorduck of computer programs are not “real” intelligence. phenomenon particularly eloquently, saying: “ It’s part of the history of the field of artificial intelligence that every time somebody figured out how to make a computer do 14 problems – there was a chorus of critics to say, ‘that’s not thinking’.” something – play good checkers, solve simple but relatively informal done yet” haven’t “Intelligence is whatever machines 15 1970 Larry Tesler, – 10 The New Physics of Financial Services |

12 Context and approach | Definition(s) of AI really mean when they say ‘artificial intelligence’? So what do people While no one can agree on they clearly mean something when they use the term AI what is, hen business people talk about AI, they typically are - defined school of computer science, rather they are W not talking about a particular technical approach or a well e a talking about a set of capabilities that allows them to run their business in a new way. At their core, these capabilities ar lmost always: A suite of technologies, enabled by adaptive predictive power and exhibiting some degree of autonomous learning, that have ma dramatic advances in our de ability to use machines to automate and enhance: Pattern detection in data Recognize ( ir )regularities Foresight Determine the probability of future events Customization Generate rules from specific profiles and apply general data to optimize outcomes Decision - making Generate rules from general data and apply specific profiles against those rules Interaction through digital or analogue mediums Communicate with humans driven technology use a combination of the above automations and enhancements - of AI Many applications 11 The New Physics of Financial Services |

13 AI and other t echnologies | Context and approach its – capabilities It is important to understand that AI does not exist in a vacuum will be intertwined with the development of all other technological innovations Emerging technologies are mutually reinforcing, and the abilities of any one new technology are influenced by its interactions with other technologies Focus on AI alone is not sufficient to understand the myriad ways in which it Several examples of mutually reinforcing interactions could be used within financial institutions. Much like other disruptive technologies, AI is not a panacea, and must be understood within the context of all other technologies that will affect how businesses operate. any one technology will increase the capabilities of all other Advances in technologies that interact with it. For example: Block - offers a source of immutable data that does not require Blockchain • centralized verification, which could be critical for identity management chain omputing holds the potential to break many encryption c • While quantum methods today, it may also bring new and even stronger techniques that will lockchain b make more secure Advanced and different computation methods provided by quantum • Decision - making for c omputing will allow AI to tackle new problems that were previously smart contracts incalculable Quantum AI will enable increasingly complex and automated smart contracts to be • omputational New c Data blockchain cases to enter the mainstream - use executed, allowing more c apabilities • both the data storage and the processing power computing will provide Cloud necessary to train new AI models, in turn making cloud infrastructure a critical part of organizations AI Cloud The potential list of interactions is endless, and will continue to develop and grow as these technologies mature and new disruptive technologies come to Processing fruition. power 12 The New Physics of Financial Services |

14 | Report purpose approach Context and Initial conversations confirmed there is an acute need to improve our understanding of the strategic implications of the suite of technologies we call AI financial services Our identification of a gap in research on the strategic implications of the future of AI in AI in financial s ervices , this work has focused mostly on observing and reporting near - term trends, or detailing While there is a huge volume of work investigating the role of technical requirements. Covered by a wealth of technical and operation research and research efforts Explored by numerous white papers Technical foundations reporting Trend Gap in today’s discourse implications Strategic There are a vast number of reports tracking the Computer science literature and research by leading emergence and development of AI and reporting on institutions have produced a wealth of material on emerging - use cases how to implement and optimize AI solutions ? that There is a lack of content explores changes to the shape and structure of financial institutions the competitive nature of financial markets and from the increased use of AI resulting 13 The New Physics of Financial Services |

15 methodology Context and approach | Research of The World Economic Forum, with support from Deloitte, has conducted one the world’s largest studies into the impact of AI in financial services  Seven that brought together global workshops  with 200+ subject matter expert interviews  Ten months of extensive research stakeholders from different backgrounds leaders across incumbents and innovators* innovators & academics orking with leading incumbents... ...and with leading W Hosting interactive discussions in financial capitals around the world Sydney, Australia New York, USA Zurich, Switzerland Hong Kong, SAR San Francisco, USA Davos, Switzerland London, UK *See page 155 for full list of contributors 14 The New Physics of Financial Services |

16 Context and approach | Report s cope regulators view of and This report will provide policy - makers with a , executives models and competitive dynamics in financial services AI’s impact on operating This report will... This report will NOT... how the capabilities of any particular AI technical details of Delve into the − − Describe how existing AI capabilities are changing the operating models of technology works financial institutions Explore how AI is shifting the strategic priorities and − competitive dynamics − Provide recommendations for how any one institution should optimize of financial services their competitive positioning by using AI - and longer - term challenges that create regulatory and public Raise near − Detail implementation strategies for pursuing potential new opportunities − policy uncertainties This report will help... the direction for their institutions, based on a heightened awareness Strategic decision - makers chart − of how the basis of competition is changing − Regulators understand the challenges they face and what responses are necessary to protect consumers and institutions alike responses in financial services and craft − Policy - makers weigh the benefits and threats of new technologies 15 The New Physics of Financial Services |

17 Context and | References approach References / Anyoha, Rockwell. “The History of Artificial http ://sitn.hms.harvard.edu/flash/2017/history - artificial - intelligence Retrieved from 1. Intelligence”, as of 28 August 2017. “History in the Computing Curriculum”. Retrieved from https ://web.archive.org/web/20110719211222/http:// www.hofstra.edu/pdf/CompHist_9812tla6.PDF 2. Osborne, Adam (1980). “An Introduction to Microcomputers. Volume 1: Basic Concepts (2nd ed .)”. Berkeley, California: Osborne - McGraw Hill. ISBN 3. - 931988 - 34 - 9. 0 4. https ://bits.blogs.nytimes.com/2013/02/01/the - origins - of Lohr, Steve. “The Origins of ‘Big Data’: An Etymological Detective Story”, as of 1 February 2013. Retrieved from big - data - an - - etymological - detective - story / 5. “About AWS”. Retrieved from https ://aws.amazon.com/about - aws / Amazon. Evans, Dave. “The Internet of Things”, as of April 2011. Cisco. Retrieved from https :// www.cisco.com/c/dam/en_us/about/ac79/docs/innov/IoT_IBSG_0411FINAL.pdf 6. 7. Orenstein, David. “Stanford Team’s W in in Robot C ar R ace N ets $2 Million Prize”, as of 11 October 2005. Retrieved from https :// news.stanford.edu/news/2005/october12/stanleyfinish - 100905.html - Somers, James. “Is AI Riding a One - Trick Pony?”, as of 29 September 2017. MIT Technology Review. Retrieved from https ://www.technologyreview.com/s/608911/is - ai 8. riding - a - one - trick - pony / report/2016/06/25/from 9. “From Not W orking to Neural Networking”, as of 25 June 2016. The Economist. Retrieved from https :// www.economist.com/special - - - not working - to - neural - networking 10. M oves, Alphago and Lee Sedol R edefined the Future”, as of 16 March 2016. Retrieved from https ://www.wired.com/2016/03/two - moves - alphago - lee - sedol - Metz, Cade. “In Two - / redefined future “Roundup of Machine Learning Forecasts and Market Estimates, 2018”, as of 18 February 2018. Forbes. Retrieved from 11. and - https://www.forbes.com/sites/louiscolumbus/2018/02/18/roundup machine - learning - forecasts - of - market - estimates - 2018/#4adcf9c02 225 - 12. “Banks A re S pending B illions of Dollars on AI to Give T heir C ustomers an Amazon - Like Experience”, as of 14 December 2017. KPMG. Retrieved from http :// www.kpmg - institutes.com/institutes/advisory institute/events/2017/12/podcast - banks - spending - ai.html - 13. “Realizing the Full V alue of AI in Banking”, as of 4 April, 2018. Accenture. Retrieved from https ://www.accenture.com/us - en/insights/banking/future - workforce - banking - survey 14. McCorduck, Pamela. “Machines Who Think: A Personal Inquiry into the History and Prospects of Artificial Intelligence”. 2nd ed . N atick, Mass: A.K. Peters, 2004. http://www.pamelamc.com/html/machines_who_think.html from http://www.nomodes.com/Larry_Tesler_Consulting/Adages_and_Coinages.html 15. “CV: Adages & Coinages”. Nomodes. Retrieved 16 The New Physics of Financial Services |

18 The new p hysics of financial s ervices Executive summary

19 | The new physics of financial services summary Executive AI is enabling financial institutions of every kind to drive new efficiencies and deliver new kinds of value... Key financial o s e ervices nabled by AI pportunities Deposits and lending Improve client advisory by Automate and augment business integrating into data streams for - credit decision making specific opportunities that could - The diagram on the right illustrates a wide array of sector opportunity discovery be enabled by the deployment of AI in financial institutions . These opportunities (and the way institutions choose to pursue them) are the foundation that drives the key findings of Use proxy data to insure new risk categories this report. Predict defaults with greater accuracy Develop Process claims modularized instantly In this diagram we have arranged these opportunities along a spectrum of five classes of policies Improve trade speed and price using dynamic execution methods on Offer tailored, always - strategies. centre of the from the Moving diagram to the periphery, the five strategy types experiences across channels Introduce new pricing and payment models Deploy new order types to protect range from conservative improvements to bold bets on new capabilities, highlighted below. investors from risks proactively Reduce fraud using new tools and new data Provide personalized, practicable - Deploy holistic market advice in real time surveillance services Provide predictive analytics to clients that help them better understand their risk Increase the capabilities of Miniaturize unsecured Develop macroeconomic Doing the same Doing something sales agents and advisers lending to be use specific - indicators using internal data Advise clients on prevention strategies Automate alert triage, to lower their risk exposures thing, better radically different investigation and reporting Triage and grade claims to increase adjudicator efficiency Offer insights on market Provide Just - In Automate reconciliation and structure and risk Develop modern, mobile - first Increase capital efficiency through better risk Time Lending incident reporting to improve insurance offerings - modelling and real time risk monitoring service quality and cut costs Sell internal analytics Improve underwriting, pricing Improve scale efficiencies to enter - Integrate post trade workflows to capabilities “as a service” efficiency and accuracy new markets - through processing achieve straight B D E A C D E C B A - trade analysis, Automate pre Act as the ultimate personal due diligence and administration Automate compliance and reporting - time pre Develop real - and post - shopper for customers management solutions - trade risk Ubiquitous Tailored Smarter New value Leaner, faster Offer seamless Drive loyalty by offering account setup and Automate investment bespoke incentives and Offer prediction “as a propositions products & presence decision perations o - customer acquisition monitoring and reporting Use broader and better data to rewards service” to merchants develop predictive risk models that drive capital savings advice making Create an advisory capability for Digitize customer Control ballooning Use predictive models to improve deal macroeconomic trends servicing - back office identification, pairing and sales activities compliance costs Compete to become a provider of invisible payments infrastructure Improve deal identification, pairing and sales activities Offer outcome - Equip advisers with Holistically Achieve better investment based portfolio highly personalized understand investor Deploy real - time For convenience, we divide the financial system into six sectors that together encompass performance by using modelling insights preferences in real surveillance capabilities new data in opaque time markets (more or less) the entirety of financial services. These are: Use existing but unused platforms Increase detection for distribution precision to eliminate false positives • Deposits and lending Establish passive products Share more detailed economic insights that track new datasets • Insurance • Payments Analyse vast quantities of Mimic advanced strategies Continuously source new data at scale while controlling costs and exclusive datasets • Investment management • Capital markets Find new and Enable users to Develop unique effectively manage unique correlations strategies and new investment products between datasets their investments • Market infrastructure Investment management 18 The New Physics of Financial Services |

20 Executive summary | The new physics of financial services ...but focusing exclusively on the capabilities that AI offers risks missing the of financial services occurring in the physics fundamental shift that is The first - movers in the deployment of AI are able to compound their lead The competitive dynamics of the data advantages to the Accelerating early offices, benefit of both front and back deeply financial ecosystem are being influences firms’ strategic approach to alliances, upended infrastructure and talent The operating models of financial Driving the formation of bifurcated markets institutions are being fundamentally where scale and agility win at the expense of - mid scale players reshaped bonds that have historically held he T Making financial institutions more specialized, and dependent on the networked leaner, highly together financial institutions are players a variety of technology capabilities of weakening Creating emerging of gravity where centres new are being combined and established capabilities ways in unexpected 19 The New Physics of Financial Services |

21 | The new physics of financial services summary Executive Yet shared prosperity in this future is not guaranteed, and requires deeper cross - prevails today ecosystem collaboration than that which Workforce engagement is critical to the large scale deployment of AI in - financial institutions Institutions must balance their While AI is often seen as a substitute for human competitive impulses against talent, establishing a workforce that views the implementation of AI as an opportunity will be collaborative opportunities critical for anything but the most marginal Time, energy and resources must be business transformations. Achieving this will AI offers financial ability to institutions the committed to resolve outstanding require an honest and collaborative relationship to solve fundamentally a host of shared problems regulatory uncertainties between an institution’s workforce and leadership but – that plague the industry and its customers together to build shared come only if they can AI development must serve the needs The deployment of AI across financial services solutions raises challenging questions about the protection of customers and remain in the best of both consumer interests and the stability of the interest of society financial system – addressing these challenges The deployment of AI should enable a fairer, engagement across private - public will demand more accessible and more stable financial and beyond institutions regulators, financial centric approach to - system. Maintaining a human the deployment of AI will be critical to ensuring it serves the interests of both individuals and society at large 20 The New Physics of Financial Services |

22 Executive | Reading guide summary What is in this report? questions: will address the following that This report is split into four sections 3 4 1 2 What are the opportunities How might the continued How is AI being deployed by How will AI reshape financial and challenges for evolution of AI affect the of financial services? sectors services? implementing AI in financial future of financial services? explorations Sector Key f indings services today? the scenarios for if?” Selected “what Page 85 22 Page of AI for implications term - long Opportunities, challenges and services financial i mplications societal broader 142 Page 55 Page Our exploration of several potential AI applications across Our predictions for the evolution of financial of Our understanding perspective of the near Our term impact of AI - disruptive scenarios that may proliferate in sectors services as a result of institutions pursuing services, highlighting key in financial on customers, financial institutions, regulators the longer term strategies enabled - AI AI - enabled strategies for institutions and society at large 21 The New Physics of Financial Services |

23 How will AI reshape financial services? Key findings

24 AI is changing the physics of financial services ... weakening the bonds that have historically held together financial institutions, while creating new centres of gravity where new and old capabilities are being combined in unexpected ways

25 Key findings | Shifting competencies in AI is altering the attributes necessary to build a successful business financial services offs will create a new wave of transformation across the global financial services industry... The resolution of previous trade - Dominant institutions in the past In the future, these institutions will be built on... were built on... Scale of assets data Scale of Economies of scale presented a cost As AI drives operational efficiency, economies of scale advantage alone will not sustain cost advantages production Mass Tailored experiences Physical footprint and standardized products AI allows the scaled distribution of highly customized - drove cost effective revenue growth products and personalized interactions Optimization and matching Exclusivity of relationships Ability to have direct access to many markets and Connections are digitized, increasing the importance of optimizing the best fit between parties connections to investors was a critical differentiator witching c osts s enefits b retention High High Continuously improving product performance to offer superior to High barriers switching providers drove customer outcomes and new value will keep clients engaged customer retention ngenuity erformance Dependence on human i Value of augmented p Processes scaled through The interplay of strengths across technology additional and functional training and talent amplifies performance labour reaching consequences for the This shift will have - up of financial services, placing legacy business models under pressure far - make are built around these new attributes from those whose businesses - Card Scenarios Key Findings Cross - Sector Impact Sector Explorations Wild 24 The New Physics of Financial Services |

26 | Overview Key findings The changing physics of financial services will transform the financial ecosystem in the following ways Key findings : AI in financial services will... From cost c entre to Institutions will turn AI enabled operations into external services, both accelerating the rate at which these - and back Make front - office - 1 others compelling capabilities improve and avoid falling behind those capabilities to to become consumers of operations look radically entre c profit different b new for attlefield A to the bottom” As past methods of differentiation erode, AI presents an opportunity for institutions to escape a “race 2 in price competition by introducing new ways to distinguish themselves to customers customer l oyalty Future customer experiences will be centred on AI , which automates much of customers’ financial lives and improves - Self driving inance f 3 their financial outcomes for solutions Collective timeliness Collaborative solutions built on shared datasets will radically increase the accuracy, and performance of 4 competitive functions, creating mutual efficiencies in operations and improving the safety of the financial system - non roblems p shared Bifurcation of market As AI reduces search and comparison costs for customers, firm structures will be pushed to market extremes, Create major shifts in the 5 amplifying the returns for large - scale players and creating new opportunities for niche and agile innovators tructure s structure and regulation of financial markets In an ecosystem where every institution is vying for diversity of data, managing partnerships with competitors and 6 a lliances Uneasy data critical, potential competitors will be but fraught with strategic and operational risks The p ower data of - will shape the relative ability of financial and non Regulations governing the privacy and portability of data financial 7 institutions to deploy AI, thus becoming as important as traditional regulations to the competitive positioning of firms r egulators Raise critical challenges for alanced Finding a b Talent transformation will be the most challenging speed limit on institutions’ implementations of AI, putting at risk the 8 society to resolve that fail to effectively transition talent alongside technology competitive positioning of firms and geographical areas to t pproach a alent the ethical grey of principles and supervisory techniques to address examination - AI will necessitate a collaborative re 9 ilemmas d ethical New institutions’ willingness to adopt more transformative AI capabilities reduce that uncertainties areas and regulatory Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 25 The New Physics of Financial Services |

27 Key findings | Reading guide This section of the report will examine each key finding, exploring the impacts and implications of the new physics of financial services The following slides will explore each “k ey f inding ” in three parts o the realization of this future, and detailing how t t These findings describe what the landscape of financial services may look like in the future, providing early signs that poin different stakeholders might be affected 2 3 1 Description Implications and uncertainties Evidence and e xamples Early that point to the Overview of the intricacies of each indicators Showcase of how the incentive structures stakeholders different realization of this finding finding in terms of how it will shape the affect uncertainties remain future of the what and industry to be determined - Card Scenarios Key Findings Cross - Sector Impact Sector Explorations Wild 26 The New Physics of Financial Services |

28 From entre Key findings | Finding 1: entre cost c to profit c - - enabled back office processes can be improved more rapidly by offering them AI to competitors ‘as a service’ model Back - office “as a service” m odel Traditional − Institutions develop AI - driven centres of Institutions develop AI − - driven of centres Institution Institution excellence around certain processes, yet excellence around given processes, and Data and have other processes that lag behind best - rocessing p offer that process “as ” a service - in class capabilities Activity 1 Fintech External activity specialist These processes continuously learn and − − Achieving excellence across all processes rd improve using data from collective their incumbent party 3 activity External Activity 2 is challenging users, improving at a rate faster than could rd of Centre incumbent party 3 Activity 3 be achieved by any one institution , competitors will move to − In the long term excellence Operations Continuously replicate the efficient capabilities of a few This creates a defensible advantage in − improving Data p rocess the defensibility of this institutions, limiting efficiency and a sustained revenue source Data advantage rd incumbent 3 party Operations Data Data capabilities Lagging Lead capabilities nd the data dynamics of AI - The proliferation of externalized services across back office processes is a result of both the rise in modularized operations a Turning operations into service offerings Operations are modularizing as institutions look As external services continue to improve, other economic advantage to commoditize their costs provides a defensible institutions are pushed to become consumers Beyond creating revenue streams, these services bring in Cost commoditization is pushing institutions to modernize Individual institutions will not be able to compete their operations to make them more more data, which ensures AI models continuously improve interoperable pendently with inde services the efficiency of collective − − Institutions become increasingly locked into Investments in modernized operations can improve the As more institutions become users of a service and integrate − mutualized the – learn from their data algorithms their data flows, underlying services built on collective data. Individual institutions would immediate efficiency of those processes, as well as lay continuously increasing the efficiency of the overall process need significant innovation to develop algorithms that groundwork to turn those processes into service offerings comparative data deficit overcome their quickly move institutions ensures This − that to modernize based processes, are - These operations, particularly AI − based architecture, operations have the potential to increasingly built on modular and cloud − Data advantages are more sustainable than algorithmic sustained develop a - operational advantage meaning they are easy to scale to new users research is open source and new AI advantages, as competitors innovations can be mimicked by 9 8 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 7 2 3 4 5 6 1 27 The New Physics of Financial Services |

29 | Finding 1: c profit Key findings entre From cost c entre to software ‘as a service’ Existing examples of offerings provide a blueprint upon built - which a majority of AI based services in financial institutions could be those from BlackRock and Ping An, have demonstrated the success of this model such as scale service offerings, - We observe that large technology and legacy are moving to the cloud and microservice Across financial services, both new applications architectures. IDC estimates that by 2021, hyper - agile architectures will be with 80% of application mainstream, Modular microservice 1 development done on cloud platforms using microservices and cloud At the back end, cloud infrastructure makes functions. architecture at the heart of third of all IT spending in financial services and is growing at over 20% CAGR as institutions push to migrate legacy up a future banking technology 2 technology onto modern platforms. 3 The Virtuous Cycle of Data academic s cientist by the The virtuous cycle of data is a conceptual framework proposed Stanford and former Baidu chief 3 AI, in effect creating and others. It describes , Andrew Ng how institutions can set up continuously improving services using roduct Great p Facebook’s timeline and Weekly Discover Spotify’s Google a defensible offering with high switching costs. Translate, have been progressively improving over time as users interact with those algorithm are examples of services built on AI that attracts i mproves Google Adwords that start services . Competitors exhibits similar virtuous data cycles for corporate customers. from scratch Development of products with with these services will face a steep uphill battle. with the goal of competing explicit virtuous data cycles Users Data enerate g BlackRock Ping An are two examples of incumbent financial institutions that internally developed world - class services, and and decided to externalize those services to develop new revenue streams. BlackRock’s Chief Operating Offer, Rob - in - Externalization of best class 4 digit Aladdin revenue growth . Goldstein, has stated that BlackRock is committed to continuing double - Ping An’s processes by incumbents OneConnect is its internal advanced technology infrastructure, covering everything from core banking technology to advanced AI capabilities. They have transformed this technology into a service offering used by nearly 500 banks across 5 China. Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 9 2 3 4 5 6 7 8 1 28 The New Physics of Financial Services |

30 | Finding 1: entre c Key findings entre From cost c to profit will shift the competitive basis of firms back office The transformation of the the distribution of talent in the industry change towards the front office, and I m p l i c a t i o n s Market power shifts to service Reduced redundancies and Operational efficiency is removed Talent will shift from financial institutions to service as a competitive differentiator increased concentration providers providers offerings become As As financial institutions become As service provider institutions become primarily office processes become - Back consumers of capabilities, jobs will collectively reliant on a diminishing increasingly uniform across financial increasingly efficient, institutions that flow ou services as most institutions will t of financial institutions but be consume number of critical systems, flaws those services face high switching costs, allowing service providers. consume similar capabilities, forcing within those systems have a recreated in service onsequently institutions to look for new areas of providers to charge high magnified impact on the financial margins C , roles may look differentiation considerably system different U n c e r t a i n t i e s How will incumbent institutions, which How will institutions protect the aspects of third - party services Which party - that the move to third Given will regulators approve of and invested heavily competitive value of their proprietary not traditionally have services is expensive and h ow in research and development, build will that data must data in a world where champion, and outstanding operationally burdensome for most be regulatory to access centres of excellence that are and prudential concerns be shared with competitors firms, which types of service offerings resolved? minimum requirements of efficiency? attractive service offerings? ? will succeed and which will fail 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 7 2 3 4 5 6 1 29 The New Physics of Financial Services |

31 for l A new b attlefield | Finding 2: customer findings oyalty Key Historically important differentiators for attracting and retaining customers are eroding, while AI is enabling new ways to differentiate offerings and win clients Emerging Historic differentiators for attracting customers differentiators for attracting customers AI is driving a dramatic shift in how financial to financial institutions in remaining The factors that were important As it frees financial institutions from the need to make trade - offs institutions should attract and retain active customers between better service and cost, AI is giving rise to a new set of competitive are reaching equilibrium, reducing institutions’ ability to competitive factors on which financial institutions can differentiate differentiate using these factors Customization Price The ability of institutions to optimize financial outcomes Platforms are making price discovery much easier, by tailoring, recommending and advising customers - margins and creating a “winner placing pressure on better will allow them to compete on value offered all” environment for - takes the institution able to provide the lowest price The old methods of differentiation are becoming a race to the bottom as AI commoditizes these methods Capturing attention Speed The ability of financial institutions to engage users and AI and other technologies enable more and more As access data through ongoing and integrated interactions time, - products and services to become instant and real beyond financial services (e.g. offering forecasting fast services will cease to be a differentiating factor (and services to merchants, booking repairs for vehicle become table instead - stakes functionality) damage, etc.) will allow them to better retain customers AI is creating new differentiators time, At the same , driven by access to unique datasets and virtuous Developing ecosystems data; institutions must us e emerging cycles of Access technologies and secular trends to remain competitive Financial ability to bring together data from institutions’ As the move to digital distribution and servicing of multidimensional consumers, networks that include their financial services accelerates, extensive branch and to offer third corporate clients and will allow them parties needed; allowing digital broker networks are no longer differentiated advice and improve performance connections (e.g. via app) to become the norm 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 7 3 4 5 6 2 30 The New Physics of Financial Services |

32 A new Key findings | Finding 2: oyalty b attlefield for customer l to differentiate themselves by using AI to build Financial institutions are seeking new products and data ecosystems We observe that AI is shifting the basis for competition within financial services to mimic competition within the technology sector RBC’s in diversifying its digital platform have allowed it to incorporate a broader ecosystem of services. For investments based on customer data. By RBC is piloting a forecasting tool for car dealers to predict demand for vehicle purchases example, are institutions Financial solutions, RBC dealers to offer RBC lending products more car its lending offering this tool alongside gives an incentive to services offering integrated rentals on Airbnb. This frequently. Other examples, among many, include software that helps register start - ups or facilitate beyond financial products ,7 6 in the race to own customers’ data and attention. allows them to better compete Lloyds Financial institutions in the UK are revisiting their core value propositions now that open banking has come into effect. Banking Group’s transformation investment of $4.1 billion a year is positioning the company to combine banking and insurance Institutions are shifting their services, along with new API - enabled propositions, to compete in the digital world. This is supplemented with a major focus on role to become ecosystem AI capabilities to transform the customer proposition and business operations. The aim is to be an ecosystem provider and a curators 8 guardian of data” in the age of many providers. “trusted scale a massive and products to achieve services has aggressively invested in building a suite of ecosystem partners, Ping An Institutions with strong of data beyond just financial services. Through a suite of apps in finance, medicine, cars and housing, it is able to take ecosystems multidimensional 9 advantage of business. core its For example, and 300 partners to power data from over 880 million users, 70 million businesses access massive scale of data cover insurance, payment and telemedicine, Ping An is able to see the gaps in service and address them to that by offering apps and insight dramatically improving the overall quality of its offerings. improve diagnosis efficiency and accuracy – Competition between people crave engagement over exposure, and prefer brands that create experiences that are In what Google calls “Gen C”, technology companies relevant and valuable. Companies such as Google and Amazon harness this behaviour by developing products and experiences foreshadows how financial sector with – boundaries cross that the intention of using their technological advantage and scale of data to provide differentiated institutions will compete for 10 customers. and higher - quality experiences that will further attract and retain customers 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 7 3 4 5 6 2 31 The New Physics of Financial Services |

33 b for | Finding 2: A new findings attlefield Key customer l oyalty will need to radically alter the way they work and the types of Incumbents products they develop in order to compete for customers I m p l i c a t i o n s Margins will be squeezed for and a Product Large tech firms have distinct development ed detailed insight ne Institutions institutions that do not develop willingness to experiment will be into customer behaviour both advantages in attracting new critical skills for side inside and out customers new differentiators institutions financial services Institutions are slow to implement To succeed, incumbents will need to The core strategies of tech that harvest new resources and have Institutions will need to be highly companies ways of new ways of differentiating their been highly focused products face an uphill battle including technical AI skills, on capturing user attention (and data) focused on delivering what customers working, product development capabilities, maintaining margins, especially after by offering free products and actually want. This means getting to just their traditional metrics such as price and services. Financial services offered know customers beyond new datasets and cultures of speed are normalized due to experimentation and looking for opportunities finances by these players will benefit from the innovation and shelf lives day existing service - to - to improve their day technology U n c e r t a i n t i e s How will regulators As financial institutions build datasets What is the natural equilibrium of Will incumbents be successful in manage the - added services price in a platform economy and what that offering value that stretch to new industries, what complexity (e.g. in assessing suitability) created by customized are the boundaries and principles that compete with existing offerings in margins can institutions expect to different industries? product attributes ? earn without differentiation? should be obeyed regarding customer privacy? 9 8 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 7 3 4 5 6 2 32 The New Physics of Financial Services |

34 Self inance Key findings | Finding 3: f - driving experience by allowing customers’ AI can deliver a radically reimagined customer in of need moments finances to run themselves, and acting as a trusted adviser Financial advice, part of every product, is flawed vision of finance could transform the delivery of financial advice A “self - driving” advice is generic and delivered Most financial Consumers interact with the agent for Products are provided by product impersonally ( , with average based - e.g. calculator manufacturers (e.g. financial Institutions) advice and to customize their products assumptions) Complex decisions receive advice e.g.: Retirement Corporate purchase Home financing planning Reliance on subjective advice from different Users - Self customer service agents leads to suboptimal Products of perception Level driving financial outcomes a gent Routine decisions are automated e.g.: p Treasury and ayment, Bill Refinancing loans cash s low f avings g sharin Advice is based on limited information as Data product and customer information is of ineffectual within and across institutions AI enables self - driving finance in three key ways customization Empowered platforms and advice Mass Continuous optimization Algorithms operating below the level of perception will The ability to compare and switch between products and Advice will be increasingly personalized and products will automate most routine customer decisions providers is critical to managing financial decisions increasingly be bespoke thanks to the use of data Advanced recommendation engines can be used to − - provider − AI can unlock the potential of emerging multi day financial management will be automated as to - - Day − product, customize the features and price for each financial by defining customer experiences platforms - platforms emerge always algorithms on and advisory using data from a variety of sources, including customers, − - - and product in Institution algorithms are critical agnostic . example, algorithms can optimize cash flows (e.g − For - and third groups sources party matching customers with products based on price and fit as savings rates), avoid fees, monitor for better deals and , − a frequent Automated, sophisticated advice for allows data determined by switch providers when necessary proactive and personalized service that is not economical − Platforms can dominate the race for customers by providing − This will result in better customer experiences around price based customer service models - under traditional human the best financial outcomes and product 9 8 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 7 4 5 6 3 33 The New Physics of Financial Services |

35 - Key findings | Finding 3: Self driving f inance A growing number of financial institutions are applying AI to customer advice and interactions, laying the groundwork for self driving finance - driving finance are being developed and implemented by fintechs and incumbents - We observe that the necessary components for self alike In the past, personal financial management apps were restricted to describing a customer’s financial situation; they were unable Clarity Money, to provide actionable insights and recommendations. The next generation of these services (e.g MoneyLion . Next personal generation - and others) are using AI to offer mass advice and customization to help improve customers’ financial positions (e.g . refinance a financial management debt loan, consolidate credit card are being built for corporate clients (e.g., ). Similar tools or cancel certain recurring payments institutional investor dashboards.) Citi launched a mobile app that allows customers to link their accounts across providers to deliver a 360 - degree view recently Tools to consolidate 11 WeChat have emerged to enable continuous Chat platforms such as providers. all banks and across of their financial lives customers’ financial lives across different financial services verticals. interoperability optimization of financial positions, and are offering - emerged that are automating routine savings and bill Myriad activities. On the savings side, several payment new apps have Automation of savings and bill apps, account. For bill Acorns, round up payments to the nearest dollar and transfer the balance to a savings such as payment customer’s cards into one account and pay each individual bill through of a all can aggregate Tally ups such as - payments, start line a single of credit. New core banking infrastructure offerings are embedding all products into a single cloud system, allowing institutions to tre at Primacy of customization in their entire product portfolio as a single balance sheet and to enable dynamic customization and pricing. For example, Thought future banking infrastructure allowing new products to be quickly customized and “smart contracts”, core infrastructure treats products as Machine’s code. direct deployed through a wizard or for which provider platforms are increasingly looking to empower their offerings through personalized recommendations - Multi provider platforms - Multi which has found success as a lead generator for loans, , products and features are best suited to customers. Credit Karma managers becoming financial 12 experiences. tools and extend their control of customer adviser in March 2018 to build financial million raised $500 9 8 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 7 4 5 6 3 34 The New Physics of Financial Services |

36 f Key findings | Finding 3: Self - driving inance - driving finance will upend existing competitive dynamics, pushing returns to Self the customer experience owner while commoditizing all other providers I m p l i c a t i o n s The frequency of customer Product manufacturers will shape Owners of customer experiences Conduct risk will be transformed themselves around the algorithms will earn the largest margins be greatly reduced interaction will There will be a large reduction in Ownership of customers will be sticky conduct risk as sales activities will be interactions fewer lose direct Product manufacturers will far There will be customers between providers and customers as self Competition will . as access to driving agents become more - driving agents performed through self - However, staff. as opposed to sales accurate as they increasingly revolve around the customer experience is largely learn and collect misconduct does occur, it will - algorithms of self the optimizing when more data. This will allow owners of automated. H owever, the interaction points that persist will become due to the scale a much larger on be driving agents rather than targeting customer experiences to exert market connectivity of AI directly advice - share of lion’s power and accrue the customers increasingly important and profits centric available U n c e r t a i n t i e s How do we ensure that algorithm - What monetization models are During what situations will human Will it be incumbents, new entrants or necessary to ensure that the can be trusted making - driven decision large technology companies that advisers need to remain involved to manufacturers i ncentives for product and held accountable? this deliver the self meet customers’ needs driving agent? - ? Will aligned the interests to of are need diminish over time? consumers self - driving agents? and 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 7 4 5 6 3 35 The New Physics of Financial Services |

37 findings roblems Key | Finding 4: Collective solutions for shared p built on shared datasets, can enable a radically Collaborative AI tools, driven - safer and more efficient financial system Example Collective Fraud Prevention Model Traditional model Collective solution Proactive Response Institution #2 Institution #1 Institution #1 Bad actor Institution #2 Institution #3 Bad actor Institution #3 Operations Operations Operations Data Data Data Predictive AI model Collective data Lagged Lagged Lagged response response response based collective solutions present a significant opportunity to address - AI Our research suggests that core challenges of the financial system prevention and anti such as fraud Processes - There is strong potential for cross - institutional AI presents a strong mechanism to collaborate as laundering suboptimally run are money - controls the value of datasets is tremendous collaboration on these issues shared Due to data asymmetries, many activities today are run processes are rarely key differentiating These inefficient AI benefits from scale in data to deliver performance that is inefficiently and ineffectively parts capabilities, greater than the sum of its individual giving institutions flexibility over execution Many of these systematic inefficiencies are highly correlated; involve Processes that often – and flows of they would not be sacrificing strategic factors critical data Since − − large datasets nstitutions would be more likely to - process at one institution has knock the suboptimal are prime on targets for AI implementations for their differentiation, i collaborate and mutualize these effects on other institutions and across the ecosystem processes AI can recognize patterns and develop insights on threats – wide contagion that cross institutional being run at sector agnostic, of these processes are Many − boundaries − If unchecked, there is a threat of a system - institutions different levels in the value chain across at – response is critical to the resolution of these threats Timely − As institutions common utilities, new move to create different product categories frameworks that address talent, governance and technology standards will emerge 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 7 5 6 4 36 The New Physics of Financial Services |

38 solutions Key findings | Finding 4: Collective for shared p roblems based solutions are emerging that could address the numerous shared pain AI - points of the financial sector We observe that both private institutions and collectives are moving to use AI to address shared problems between 2009 and 2017 $ compliance 89% of industry totalled - with illion globally, b 342 Cumulative financial penalties for non $ b illion 342 14 in increases continued around the world expecting executives 2019. – 2017 from compliance costs Regulatory priorities stretch Increasing regulatory focus is cumulative financial penalties for capital beyond leverage and issues such as financial - adequacy requirements, and there is also increased focus on collective institutions’ straining budgets non between 2009 compliance - 13 place New regulatory requirements that emerge to address these areas will and data security. pressure on crime, privacy 2017 and operating budgets for institutions. have EarlyWarning SWIFT Collective institutions such as started developing service offerings that will leverage AI and the and collective power of data to address some of the biggest threats. SWIFT is launching a new intelligent in - network solution for - Early stage collective utilities - time monitoring, and alerting and blocking of sent payments, with daily reporting. The solution fraud control that combines real are emerging, backed by key solutions, technology - leading over time and is part of SWIFT’s commitment to develop “smarter” gets edge ML, including AI and service providers is a fraud and EarlyWarning requirements. to help its customers address their regulatory, fraud prevention and risk mitigation technology management - risk banks, started by a collective of the largest US , company that employs AI. - based transaction AI AI in using have demonstrated significant benefits Shift Technology and based such as Companies - ComplyAdvantage monitoring models have a algorithms to monitor transactions. ComplyAdvantage claims to have achieved an 84% reduction in false positive alerts for AML demonstrated advantage over 17 risk data, while Shift Technology is helping insurers fight claims fraud using AI. the status quo - time payment brings with it “real - time fraud”. There is evidence to support this contention: the Many have speculated that real time and digitized - Real 15 hanks it implemented the F UK P ayments scheme experienced and, t a to 132% increase in fraud in the year after aster transactions are presenting 16 Furthermore, e xperts have warned that the proliferation of AI double. automation, insurance fraud cases are expected to new collective risks technologies could enable new forms of cybercrime and other threats across different industries. 9 8 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 7 5 6 4 37 The New Physics of Financial Services |

39 shared Key | Finding 4: Collective solutions for findings p roblems New frameworks must emerge to enable shared accountability if collective solutions are to succeed I m p l i c a t i o n s compliance will become control must be Accountability and The safety of the financial system Efficient - Growing cyber a risks present commodity will be radically improved increasing operational challenges renegotiated Institutions mus If collective solutions succeed, real As certain processes are shifted to t develop strategies to As institutions collectivize their shared - increasing data mitigate the scanning using shared utilities, institutions will seek to time compliance services, they will full market risk of abuse and leakage of highly confidential participate on the same competitive central offload accountability to these has the potential to dramatically increase institutions’ ability to react utilities as well, while regulators information at a customer and plane – removing efficiency of will push to hold the institutions transaction level, as well as the compliance as a competitive proactively to threats and catch differentiator sharing of sensitive - increased risk accountable malicious activities with improved competitive information accuracy U n c e r t a i n t i e s What is the right ownership How will - Can cross How will liability for errors and the industry ensure border solutions be be shared compliance failures framework for collective utilities to developed given a growing leadership and investment across the financial system to overcome the ensure their between utilities, collectives and divergence in financial and data interests are aligned with barriers to collaboration? their stakeholders? regulations? individual institutions? 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 7 5 6 4 38 The New Physics of Financial Services |

40 | Finding 5: tructure s Key findings market Bifurcation of The economics of AI will push market structures to extremes, scale favouring mid firms sized players and agile innovators at the expense of - - sized mid firms Regional / Scale players Niche players Today’s m arket tructure s Consolidation of scale Proliferation of niche and Disappearance of - mid players into fewer but agile firms sized firms larger entities Future market s tructure economics operating AI will drive shifts in customer behaviour and Cause: scale players and agile innovators favour This shift will simultaneously Effect: cost products, as , allowing them to capture customers − Scale - based players have a natural cost advantage − Platforms will push customers more aggressively to switch to lower - mid price is the primary decision factor for most commoditized financial from - sized players, especially given increased price transparency in the industry that best fit their served customers by optimizing their − Optimization algorithms will help customers find niche products − Agile and niche players can capture under - - offerings to the decision and targeting unique, unmet needs unique needs, in cases where this behaviour dominates price making of algorithms (e.g. niche insurance, unique investments). This could present an opportunity for non - bank new − AI will allow new offerings to be built and scaled much more efficiently Firms can use AI . entrants (e.g. tech companies) requirements services to establish full back without massive offices regulatory e.g. meeting ( sized or the need to hire large workforces) − Mid - investment firms will struggle to make the investment necessary in AI to remain as competitive incumbent firms increasingly become AI service providers, firms that do not ; have the capacity to build similar offerings will struggle 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 7 6 5 39 The New Physics of Financial Services |

41 s findings | Finding 5: Bifurcation of market Key tructure suggest that, Economic theory and market observations as AI becomes more important, firms will be pushed to market extremes Evidence from economic theory structures to market extremes. Erik Brynjolfsson defined this phenomenon as being the result of the simultaneous creation of Economists have documented the power of information technology to drive firm 18 AI accelerates this phenomenon by - all superstar structure. - takes magnifying the impact of several key drivers: combined with a availability, long tails in product winner − - based search algorithms can better match users with what they are looking for, making more qualitative aspects of financial services easier to search and compare Search and database technology: AI By enabling continuous customization, AI helps products better meet the unique needs of customers who were previously - served under Personalization: − As AI automates back have a low office processes, new products and offerings − Making niche products cheaper to build: marginal cost, improving the economics of offering niche products - - . the lowest the cheapest products (e.g advantages in offering As search costs are minimized, scale players have inherent returns to scale: Improving − loans or cheapest insurance premiums) rate Evidence from observations Scale players such as Vanguard have aggressively pursued a low - fee offering by taking advantage of economies of advisers ) have scale. In the ETF market, automated platforms (e.g . robo ability to seamlessly optimize the increased - Market extremes are already fees, helping the scale players to win customers. At the other end of the spectrum, a new class of investments and developing in asset emerged. These are led by innovative entrepreneurs who use AI and quantitative investing to deliver funds has management differentiated return that can be scaled rapidly without substantiall y increasing their costs or profiles human capital footprint. AI solution Respondents with at least one deployed, by assets globally Firms with fewer assets are lagging behind larger in both AI and digital transformation. A survey by 3% investment firms 7% 25% DBR Research found that 48% of banks with more than $50 solution, compared to billion in assets have deployed an AI tier firms are lagging in - Mid 48% and $10 billion in % for banks with between $1 7 assets. One reason for this is that mid - tier firms have tighter billion technology investment resulting in rely on technology vendors, investment budgets and and a reduced limited internal capacity for innovation ability to move quickly. billion – $1 – billion – billion $10 $50 $50 + billion 10 billion $ 1 $ $ billion 500 million 19 Source: DBR Research 9 8 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 7 6 5 40 The New Physics of Financial Services |

42 market findings | Finding 5: Bifurcation of Key s tructure As institutions get pushed to market extremes, firm structures and core competencies at either end of the spectrum will begin to look radically different I m p l i c a t i o n s Consolidation Product shelves widen and Reduced costs of entry enable a of mid Aggressive competition will lead to layers p tier - consolidation among s cale diversify new generation of product players - tier financial institutions As mid manufacturers As more niche players enter the of favour In order to retain the become less profitable, they will New firms will be formed by recommendation algorithms, scale market and try to fill unique and become acquisition targets for scale under players to increase the size of their players will maximize their economies - innovative entrepreneurs and will be serviced needs, consumers will of scale, focus on key products and able to scale rapidly. The shape of books have access to new and different core activities in - these new firms will be radically divest from non products that better fit their financial traditional financial different order to improve price and requirements from institutions performance metrics U n c e r t a i n t i e s How will regulators react to the As operational barriers to entry are How will emerging markets react to To what degree will consolidation of assets and increased consolidation of reduced, will regulatory barriers aggressive expansion of international scale players become a cross - border firms into their domestic systems? the risk of creating new “ too big to similarly adapt to enable new phenomenon? entrants? fail” entities? 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 7 6 5 41 The New Physics of Financial Services |

43 | Finding 6: Key findings Uneasy data a lliances have winners Data partnerships losers as and emerging out of necessity will some firms are pushed to the periphery and others emerge as ecosystem hubs short - term opportunities In order to access the full benefits of AI, institutions will be pushed to enter into data partnerships for Depth of Breadth of data data Access to end u sers Machine learning is very data hungry Institutions collect more data by being closer to users Diverse datasets are critical for most AI applications use c Enables virtuous data cycles Enables more accurate models Enables more complex - ases However, these partnerships exhibit , term - long risks with the potential to create winners and losers in - lock Partnership isks r privacy Security and xperience e all Imbalance of imbalances p ower - takes - Winner Data market Institutions that become reliant on The gap in data between large techs poses ata connectivity experiences Owners of customer d - takes Customer experience is winner - Increased data flows from partnerships may and incumbents will continue to grow, , especially in platform and self and privacy risks that could security have strong market power and can pit - all become locked in unfavourable diminishing the power of asset scale providers against each other driving ecosystems break apart critical partnerships relationships will be Alliance ” “losers those relegated to the Alliance “winners will be ” periphery of the ecosystem , those who can successfully turn themselves into becoming interchangeable ecosystem hubs with other data and capital sources 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 5 7 6 42 The New Physics of Financial Services |

44 | Finding 6: Uneasy data a lliances Key findings Partnerships are proliferating across the financial services ecosystem, but only time will tell if these relationships drive sustained value We observe that data partnerships are currently in the early stages of being formed, but are not yet at the stage where tensions have major impacts JPMorgan Chase , Amazon and Berkshire Hathaway to build a health insurance alliance for The recent partnership between 20 their own employees demonstrates the power that can be through collaboration. T he alliance will use big data and unleashed Growth in partnerships increase customer engagement and improve targeted health plans (e.g . specialized medicine, technology to align incentives, between financial institutions However, many news outlets have speculated that Amazon is a challenger to smoking cessation and obesity programmes). companies technology and incumbent institutions and could potentially be a significant disruptive force in financial services. PSD2 As the UK Open Banking Standard and . Challenger banks have emerged providers go into effect, a number of new niche New entrants are expanding and third N26 rapidly expanding across Europe. These such as are – Klarna and Squirrel instance, for – providers party - more quickly as a result of data necessary to power their operations without giving incumbent institutions extract, from incumbents, the players are able to open banking reciprocal access – advantage. incumbents’ data=exclusivity essentially eroding and Pay Uber’s credit card are all examples of nominally profitable but data Apple Pay, Google - generating entries into Technology companies are companies. financial services by technology However, such affinity agreements are misaligned to shifts in the industry: data is building products to access an increasingly important input to differentiation, while the ability to generate meaningful profits from transaction revenue is and generate financial data s a result, competitive tensions between technology companies and financial institutions are likely to grow. A declining. platforms New financial services ecosystems are emerging in Asian countries, building apps on top of existing such technology Emerging Chinese financial . Taikang Life ). WeChat acts as the interoperability layer, connecting customer data with financial institutions, as WeChat (e.g - ecosystems centre on data elements at critical they grow in size, t hese the centre of these As as well as institutions to each other. platforms become sharing platforms , ecosystems while the financial services providers are interchangeable based on customer preferences. 9 8 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 5 7 6 43 The New Physics of Financial Services |

45 Uneasy Key findings | Finding 6: data a lliances Incumbents are placed in a double bind: they cannot resist entering data partnerships but these partnerships may threaten their competitive positioning I m p l i c a t i o n s Partnership development is - restrictions on data Unilateral technology c ompanies will Emerging tensions could threaten Large alliance sharing make forming partnerships becoming a critical competitive longevity become critical sources of data competency difficult and customer experience By positioning themselves as the data Due to their data advantage, large Firms that move to prevent - Effectively developing the right data critical link across the ecosystem, technology firms will be critical sharing by their customers partnerships while mitigating potential will firms can turn other participants into commoditized service providers. struggle to form the data partnerships tensions in those relationships will components of the financial value Tensions arising from this may limit allow firms to sustainably develop short term, . In the necessary to develop AI capabilities, proposition to pressure incumbents will feel unique and differentiated products, the longevity of emerging which risks leaving them alliances uncompetitive experiences and insights partner with these firms to access customers and data U n c e r t a i n t i e s regulator - How will smaller and regional banks Who will retain control of the mandated data - How will new data regulations affect How will negotiate effectively with large affect ) . open banking (e.g sharing large tech companies, and will they customer experience in partnerships the - technology sharing relative negotiating power of companies, particularly if technology companies and be held to similar data between ? already those technology companies standards as financial institutions institutions in forming partnerships? financial services firms? have major financial services partners? 9 8 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 5 7 6 44 The New Physics of Financial Services |

46 of | Finding 7: The p ower findings data r egulators Key Data regulations will have transformative impacts on the shape and structure of financial markets, particularly where they require increased data portability s egulations on market r Influence of data egulations r ata d critical Recent tructures ervices s based - loud c use to The ability can that of data the kinds The ability to use public and private cloud infrastructure, and to use be A hosted on the cloud, is critical to the development of AI applications Walled gardens ynamics d ompetitive c Impact on Regulations on cloud usage by financial institutions vary globally, with stricter restrictions in ‒ Europe with more relaxed rules have an advantage in developing new players in regions Technology ‒ capabilities . (e.g - data If large financial institutions maintain control over over sharing bilateral the terms of how they share customer data), they may prefer to strike third access to partnerships with tech companies data and AI capabilities. - party security and privacy Data This while taking to sustain their market position would allow incumbents protection regulations (e.g . GDPR) are placing new limitations and requirements Privacy and data - capabilities AI differentiated advantage of on the collection, data and storage of personal transmission on ynamics Impact d competitive Data partnerships become increasingly difficult to manage as parties are held to stricter ‒ requirements Consumers ‒ gain increasing control over their data, including control over who can access that B “ data and the right to be forgotten ” led - Platform anking Data portability and open b , UK Open Banking) Regulations in Europe (e.g . PSD2 require that incumbent institutions share ) financial data with third parties (at the request of the customer customers’ ynamics on Impact competitive d If personal financial data became hyper - portable , third parties would be able customer data and build platforms, and customer data to access incumbent - held − firms can access financial data and use it alongside a wealth of other personal technology Large would cease to be a source of competitive differentiation. In this case, data, giving them a for customers’ finances applications head start in developing new AI incumbents would likely compete on platform - based ecosystems, providing - − financial data from third Financial institutions do not have the reciprocal ability to access non commoditized products and competing against many manufacturers technology e.g. companies) parties ( 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 5 6 7 45 The New Physics of Financial Services |

47 r Key p ower of data The egulators findings | Finding 7: Global data regulations are undergoing a period of unprecedented change as governments move to adopt new rules to protect and empower citizens New data regulations cover a range of issues from privacy to data protection and portability The revised Payment Services Directive (PSD2) of the European Union came into force in January 2018, with the aim of are GDPR and PSD2 In conjunction with the General Data Protection Regulation (GDPR), this enabling more innovative payments across Europe. dramatically reshaping the means institutions have to carefully balance requirements to share data with third parties against the risk of substantial data economy European 21 penalties in cases where data is mishandled. b anking as a mandate across financial services. This push started The UK has been one of the first jurisdictions to adopt open The UK open banking standard “ older and larger banks do not have to compete hard in 2016 , with a report by the Competition and Markets Authority that found has the potential for major 22 enough for customers’ business, and smaller and newer banks find it difficult to grow”. shifts in domestic financial services While China does not have an open banking framework, the has been very conducive to f intechs and existing regulatory regime companies and incumbent institutions third - party providers. Proliferation of APIs (both public and private) between technology Emerging Chinese ecosystems (e.g . WeChat and Alipay) have allowed these platforms to become interoperability layers to facilitate the flow of data across platforms on data centre 23 institutions. governments are considering radical changes to their parts of the world, Across different data - openness regimes. Australia, o Singapore, Canada often anking regulatory model, b pen and Iran, among others, are actively considering some form of the Globally, other countries are 24 ,25 by the EU and the UK. mirroring often extend beyond financial services and the steps taken These data regulations affect considering new regulations many different sector open - Corruption Working Group has identified cross - the G20’s Anti industries collectively. For example, 26 integrity. transparency and sector - public data as a priority to advance sharing alliances are more mandated, with individual banks building bilateral In the United States, data - ad hoc than relationships with data aggregators. Regulators have not signalled an intention to implement frameworks similar to the UK and US regulators are largely silent 27 EU. Congress However, the US the has been listening to testimony from large technology companies such as Facebook, on data regulations 28 Google and Twitter on the topic of privacy and data security, which could lead to the emergence of new rules. 9 8 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 5 6 7 46 The New Physics of Financial Services |

48 of data The p ower | Finding 7: findings r egulators Key The evolution of data regulations will be the critical driver in determining the roles and relative positioning of different players in financial services I m p l i c a t i o n s stems will be sy can thrive by using AI to Incumbents must play an active Digital identity Data regulations formulated in the Fintechs coming years will have long - develop unique offerings and role in shaping data regulation if critical to managing personal data by they are to remain competitive flows on financial markets lasting effects using open banking to access data As consumers gain increased control Increased requirements Wide - reaching, cross for data - In many jurisdictions, data regulations sector data used, are still being developed. In the over how their data is data incumbents’ portability will erode regulations affecting entire economies they will coming years, these regulations will need a consolidated point of control will determine whether incumbents to fintechs advantage, allowing solidify and financial markets will be to easily manage consent and can access the necessary external compete more effectively for scale of authorization; this is likely to be a shaped by those regulations for the data to continue to own the customer data ID system experience foreseeable digital future U n c e r t a i n t i e s - What norms will develop regarding What form will new open banking and Will it be possible to effectively trans How will recent consumer concerns locate AI models developed in data privacy rules take in more , Canada international data flows, and how will regarding improper data usage and permissive data regimes to less Australia and other countries, and - cross divergent domestic rules affect W estern sharing by large tech firms in border permissive ones, and what impact data flows? how will financial institutions be countries be resolved? with this have on competitive affected? dynamics? 8 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 5 6 7 47 The New Physics of Financial Services |

49 Finding a balanced approach to talent | Finding 8: findings Key priorities of financial institutions are mismatched with the talent near The - term term - longer on a required to deliver vision for business transformation - al to long tic Transforming talent is critical to unlocking the most impactful applications of AI, making investments in talent strategy cri term competitiveness Financial institutions must - term focus on optimizing current An outsized near An insufficient long - term focus on accelerating achieve a balance between environments business transformation the rapid deployment of new technologies and the Using AI to optimize current activities centres contact . (e.g – Net new opportunities for talent will be created by shifts to – development of a talent compliance functions) offers institutions immediate and new business models and competitive dynamics ecosystem that is fit for the benefits from small scale investments and does not require a - Success in tomorrow’s business environment will be – more transformative changes clear vision of the future of the institution predicated on talent strategies and capabilities that are facing the institution – However, these optimization projects drive value primarily distinctly different from those that exist today (e.g. roles, through headcount reduction – reducing jobs faster than new culture, rewards) opportunities are created – Financial institutions that fail to evolve talent strategies If – talent erodes without a clear vision of the institution’s future - fill to - alongside business transformation risk creating hard it risks creating unnecessary stumbling blocks for financial in their talent profiles, due to an inability to attract, train gaps institutions as they strive to transform and retain people with the skills and capabilities that are complementary to machines talent and technology is creating significant roadblocks to the transformational agendas of firms: An imbalanced focus on Leadership is not aligned to the potential and ear n - continue to reward a focus on Incentives Organization structures remain rooted in traditional pace of innovation term results ways of working of financial services is future Uncertainty about the – Methods of are competing forces – reward and recognition – Incumbent environments of financial institutions remain confounding views on what talent will be required, and how that are working against longer term work efforts - hierarchical, with expansive spans and layers different capabilities can be secured – Traditional structures of employment, including work Attempts at structural change have weeded out duplication, but – – Financial executives know they need to change, but are arrangements, compensation structures and retention have yet to reshape the operations of financial institutions impacts or timing of when that not aligned on the areas, strategies, are hindering financial institutions’ ability to – A vision for structural change needs to lead the articulation of change will take place businesses by diverting forces to near their change term - new ways of working, and subsequent roles that are required performance metrics 7 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 5 6 8 48 The New Physics of Financial Services |

50 Key findings | Finding 8: Finding a balanced approach to talent Financial institutions need help to conceptualize the union of talent and technology as they aspire to move forward and achieve AI - driven growth Evidence from economic theory th observes that the current of transformation – often dubbed the Fourth I ndustrial R evolution – bears a strong resemblance to the Industrial Allen evolution of the 19 R Economic historian wave Robert C. “Engel's Pause”, In phenomenon where a new technology results in a worsening of circumstance followed by increased century. particular, the reaction of talent to increased use of AI is reminiscent of a 29 - evaluate their understanding of, and strategies for, both talent and technology transformations : prosperity. This is highlighted in a number of examples and reinforces the need for institutions to re , there is risk that unemployment rates rise: As AI digitizes and automates routine roles, there will be a net displacement of talent unable to rejoin the workforce with thei r e xisting − In the immediate term skills; the initial rise in unemployment will create the perception that AI is worsening society rather than improving it In the near term, will drive growth: For society to overcome these harms and embrace AI, the technology must be developed to a level where it is pervasive. Although this may be shift in talent − a innovation and longer - term thinking, within and outside formal institutions will need to be a shift in priorities to encourage there technologically feasible, be society and business will be redesigned: The benefits of AI will − maximized only if societal structures and processes adapt to support new ways of Over the longer term, relationships between 29 while working to address challenges , institutions and social structures will and enabling AI innovations, - driven growth By aligning with working. likely see prosperity that justifies AI term - longer in a investment - highlights historic under We observe that the current situation in financial services vision for talent and technology often thousands of people are , where there office back In an effort to optimize processes, many institutions have started in the - paper processing customer bank, European ne o requests or tackling reporting needs. At where 70% of applications were 30 based on processing , the time spent by staff and automation. digitization forms was reduced by 70% through Product improvements have been the core focus of Group are the front and back office, such as Mizuho Financial Institutions developing AI tools to increase efficiency across both financial institutions 31 which they say could replace 19,000 staff (around a third of their staff) and close 30 branches ontact centres are an area of C . are becoming disrupted by new services such as Amazon Connect, which provides cloud driven based, AI - focus. These - 32 replace the contact centre and the sizeable real estate, FTE and management associated with it that solutions . ranging agreement that, due to under - investment and lack of foresight on technological change, financial There is wide - institutions face a serious mismatch between the skills and capabilities of their current workforce and those that will be As an industry, financial incumbents have begun prioritizing efforts to reskill, these efforts required to remain competitive in coming years. Though services is playing “catch up” - , 34, 35 33 (e.g. Scotiabank, BNP Paribas, BlackRock). appear broad in their approach, though foundational steps are underway to create a talent pool and culture that are ready for Beyond reskilling, financial institutions must further reshape the internal culture of their organizations to attract and ret ain change - - - people with sought and control cultures that trap after skills and capabilities. This means moving away from the command - centred environments that embrace the future of work. individuals in narrow roles, with little autonomy, and towards employee 7 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 5 6 8 49 The New Physics of Financial Services |

51 Key findings | Finding 8: Finding a balanced approach to talent AI is giving rise to a new discipline with regard to how financial institutions plan for their future talent and technology needs I m p l i c a t i o n s Institutions face a prolonged skills Investment in talent is a critical Resistance to change will lead to a People management will become a number of false starts deficit competitive advantage enabler of AI Financial institutions that are able to The most significant roadblock to Efforts to reskill are lacking a clear Talent strategies need to be change will be created by people due developed within the context of a proactively create new talent view on the roles and responsibilities effort and defined strategy and target operating ” and how to insufficient that institutions “runway experiences through the execution of (time, need today, , evolved policies demand for skills will change over the investment) being prioritized at the model, allowing the evolution of talent processes and – lead in their ability to structures will onset of transformations to establish to be viewed as a strategic enabler The lack of vision longer term. and understanding, effectively execute, and accelerate, enacting change and from strategy puts institutions coordinated continued and in - buy risk of dependency to earning a return on motivation for commitment at business transformations a prolonged skills deficit other investments U n c e r t a i n t i e s profiles What are the specific talent How can institutions balance the How can financial institutions accelerate What should the role of government learning today domain expertise they need be during this period of uncertainty that financial services will need to transformations when training, and as new talent economies unfold? evolve and perform within new with the quintessentially human and adapting takes place at human speed? business models? capabilities they will need over the longer term? 7 9 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 5 6 8 50 The New Physics of Financial Services |

52 ilemmas | Finding 9: New ethical d findings Key Mitigating the social and economic risks of AI in financial services will require collaboration multistakeholder development Global communities have a joint interest in mitigating the risks and harms of rapid technological p rotection and the Consumer Employment and human Experience of other regional Global and s ystem financial Safety of the i ndustries p ublic c apital i rowth e conomic g nterest AI enables the dramatic AI creates new opportunities to AI expands the reach of effective driven connections - AI makes data AI provides new opportunities to across industries and borders to augment performance by making - decision through the more efficiently and effectively simplification of processes, deliver value in differentiated improving the speed, democratization of financial quality improving the productivity of and combat bad actors through ways... ... ... collective action... cost of commerce... advice workers within redesigned roles but... but... but... but... but... might continue to subject is will reduce the need for labour up the industry to broader opens susceptible to creating has the potential to polarize global risks of contagion as AI demands segments of the population to of communities as competition across routine tasks, leaving some excessive concentrations without the required , ” unfair and inequitable exclusions ost market power and driving income around AI development becomes a ”l workers an increasing interconnectedness inequality across domestic and cross - border from products or services certain skills and capabilities for new roles point of regional conflict systems being are too great to be While the potential benefits of AI will be striking, its potential risks to societal and economic well - left unaddressed 8 Key Findings Cross - Sector Impact Sector Explorations Wild - Card Scenarios 1 2 3 4 5 6 7 9 51 The New Physics of Financial Services |

53 Key findings | Conclusion Responding to the new physics of financial services will require institutions to maintain a challenging balance between competition and collaboration ervices s The growing role of AI in shaping the future of financial will require financial institutions to simultaneously... Collaborate with many stakeholders Be first and b est in the deployment of AI requires are able to Because those institutions that Because unlocking the full potential of AI establish an early lead in using AI as a competitive an extensive network of partnerships and only differentiator will be rewarded by collective efforts by financial institutions, virtuous alongside regulators and the broader public cycles that compound their advantages feedback up sector, can ensure that the expanded use of AI in and leave second movers struggling to catch finance benefits society as a whole Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 52 The New Physics of Financial Services |

54 Key | References findings References “IDC FutureScape: Worldwide Cloud 2018 IDC . Retrieved from https :// www.idc.com/research/viewtoc.jsp?containerId=US42014717 1. Predictions”, as of October 2017. Public Cloud Services Spending Forecast to Reach $160 Billion This Year, According to . Retrieved from IDC “Worldwide 2. IDC”, as of 18 January 2018. www.idc.com/getdoc.jsp?containerId=prUS43511618 :// https From the lecture “Artificial Intelligence Is the New Electricity ”. https:// www.youtube.com/watch?v=21EiKfQYZXc 3. “BlackRock Investor Day”. BlackRock Inc. Retrieved from https :// 4. event.webcasts.com/starthere.jsp?ei=1197180&tp_key=4a4a9d4d3f 5. OneConnect Fintech Unit”. Financial Times . Retrieved from https://www.ft.com/content/28f7fd06 - 332d - 11e8 - ac48 - “Ping An Eyes $2bn Listing of 10c6fdc22f03 6. K nows Y ou “Big Tech etter than Your B ank Does, and that Worries RBC’s CEO Dave McKay”, as of 6 April 2018. Financial Post . Retrieved from B http :// business.financialpost.com/business/big - tech - knows - you - better - than - your - bank - does - and - that - worries - rbcs - ceo - dave - mckay - 7. Warns on Big Tech Threat to Banking”, as of 14 June 2018. Fintech Collective. Retrieved from http ://news.fintech.io/post/102exfi/royal - bank - of - canada - warns Bank of Canada “Royal on big - tech - threat - to - banking - 8. “Strategic Update”, as of 21 February 2018. Lloyds . Retrieved from http :// www.lloydsbankinggroup.com/globalassets/documents/investors/2018/2018_lbg_strategic_update_presentation.pdf 9. “ 从 平安到平台:科技 创新赢未 来 陈 ”, as of 20 November 2017. Pingan . Retrieved from http ://www.pingan.com/app_upload/images/info/upload/5677f5d7 - 409a - 41b6 - a229 - 9e15eabec596.pdf - “The Engagement Project: Connecting With Your Consumer in the Participation Age”, as of May 2013. Think with Google . Retrieved from https ://www.thinkwithgoogle.com/consumer 10. insights/engagement - project - new - normal / “Citibank 11. ://www.citigroup.com/citi/news/2018/180326b.htm Announces National Digital Banking to Serve Clients Across the U.S.”, as of 26 March 2018. CITI . Retrieved from https 12. “Silver Is Buying a $500M Stake in Credit Karma in a Massive S econdary Round”, as of 28 March 2018. Techcrunch . https ://techcrunch.com/2018/03/28/silver - lake - is - buying - a - Lake - in - in - credit - karma - 500m - a - massive - secondary - round / stake - “Global Bite”, as of 22 February 2018. BCG . Retrieved from https://www.bcg.com/d/press/22february2018 - global - risk 13. future - Banking Recovery Stalls, as Risk and Regulatory Costs proofing - bank - agenda - 185132 14. “Opinions on Global F inancial S ervices R egulation and Industry D evelopments for the Year A head”. Duff&Phelps. Retrieved from https://www.duffandphelps.com/ - /media/assets/pdfs/publications/compliance - - regulatory - consulting/2017 - global - regulatory - outlook - viewpoint.ashx and - “Risks in Faster Payments May 2016”. . Retrieved from https://www.frbatlanta.org/ - /media/Documents/rprf/rprf_pubs/2016/risks Federal Reserve Bank of Atlanta in - faster - payments.pdf 15. 16. “Insurance F raud C ases E xpected to Double with the Rise of Automation”, as of 15 January 2018. Medium . Retrieved from https ://medium.com/@ arnaud.grapinet/insurance - fraud - automation cases expected - to - double - with - the - rise - of - - - ce6e700ccc2a https 17. - driven Risk Data & Next Generation Technology”. ComplyAdvantage . Retrieved from AI :// complyadvantage.com/wp - content/uploads/2017/12/20171211USComplyAdvantageOverivewBrochure - 1.pdf Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 53 The New Physics of Financial Services |

55 Key | References findings References “Long Tails Versus Superstars: The Effect of IT on Product Variety and Sales Concentration SSRN . Retrieved from https ://ssrn.com/abstract=1676368 18. Patterns”, as of 1 July 2014. Survival in Banking Hinges on Artificial The Financial Brand . Retrieved from https ://thefinancialbrand.com/67890 / “Competitive Intelligence”. 19. Dimon Lays Out JPMorgan - Amazon - Berkshire Health Care Priorities”, as of 4 May 2018. Investors Business Daily . 20. https ://www.investors.com/news/jamie - dimon - “Jamie Retrieved from - - amazon - chase - venture / jpmorgan berkshire “European Parliament Adopts European Commission Proposal 21. Create S afer and More I nnovative European Payments”, as of 8 October 2015. European Commission . Retrieved from to http://europa.eu/rapid/press release_IP - 15 - 5792_en.htm?locale=en - “Background / Banking”. Open Banking . Retrieved from https ://www.openbanking.org.uk/about - us 22. to Open https “Open 23. in Asia Pacific”, as of 6 February 2018. Herbert Smith Freehills . Retrieved from Banking :// www.herbertsmithfreehills.com/latest - thinking/open - banking - Developments developments - in - asia - pacific data 24. our’ to Open B anking D ata by July 2019”. ZDNet . Retrieved from https ://www.zdnet.com/article/australia - to - force - big - four - to - open - banking - F - by - july - 2019/ “Australia to Force ‘Big 25. M erits of ‘Open B anking,’ a Catalyst for F intech”. Financial Post . “Federal Budget: Ottawa to Study http ://business.financialpost.com/news/fp - street/ottawa - to - study - Retrieved from merits - of - open - banking - a - catalyst - for - fintech . 26. - Corruption Working Group”. G20 Anti Retrieved from https :// www.g20.org/sites/default/files/media/g20_acwg_interim_report.pdf “G20 27. “Why Do M ost U.S. Banks S hut the Door on ‘Open Banking’?”, as of 20 December 2017. American Banker . Retrieved from https://www.americanbanker.com/list/why - do - most - us - banks - - shut - the psd2 door - on - - style - open - banking 28. “Facebook, Twitter and Google CEOs for a Hearing on Data P rivacy by Congressional Committee on 10 April”, as of 27 March 2018. Firstpost . Retrieved from Summoned www.firstpost.com/tech/world/facebook - twitter - and - google - ceos - summoned - for - a - hearing - on - data - privacy - by - congressional - committee https:// on - 10 - april - 4407165.html - 29. “Engels Pause; Technical Change, Capital ccumulation, and Inequality in the British Industrial Revolution”, as of 8 February 2008. ELSEVIER. Retrieved from A ://www.nuff.ox.ac.uk/Users/Allen/engelspause.pdf http “Automating the Bank’s Back O ffice ”. McKinsey&Company . Retrieved https ://www.mckinsey.com/business - functions/digital - mckinsey/our - 30. - the - banks - back - office insights/automating 31. “Japan bank Mizuho eyes cutting 19,000 jobs worldwide, replacing clerical jobs with AI: Reports”, as of 30 October 2017. The Straits Times. https://www.straitstimes.com/asia/east - asia/japan - bank - mizuho - eyes - cutting - 19000 - jobs - worldwide - replacing - clerical - jobs - with - ai 32. “ Launches Amazon Connect, Productizes Amazon’s In - house C ontact C enter Software”, as of 28 March 2017. Techcrunch . Retrieved from AWS - connect/ - https://techcrunch.com/2017/03/28/aws amazon for the “Scotiabank $250M to Help R e - skill 33. mployees Invests Digital Economy”, as of 10 April 2018. BNN Bloomberg . Retrieved from https://www.bnnbloomberg.ca/scotiabank - invests - E 250m - to - help - re - skill - employees - for - the - digital - economy - 1.1052470 - 34. Spend €3bn on Digital T ransformation”. Financial Times . Retrieved from https://www.ft.com/content/c57ac5ca - ec9b - 11e6 “BNP Paribas to ba01 - 119a44939bb6 America’s Skills 35. Helping C lose “How CEOs are Gap”. Business Roundtable . Retrieved from http:// businessroundtable.org/sites/default/files/immigration_reports/BRT%20Work%20in%20Progress_0.pdf Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 54 The New Physics of Financial Services |

56 What are the opportunities and challenges for implementing AI in financial services today? O pportunities, challenges and broader societal i mplications

57 i Opportunities, challenges, societal mplications | Summary opportunities for financial institutions, but only if internal significant AI presents challenges can be overcome and societal implications effectively managed Path to AI implementation Opportunities near term , financial institutions can achieve substantial value by taking In the AI to pursue new competitive strategies across their value chains advantage of Challenges executing AI strategies will require significant effort to address However, successfully and regulation talent challenges related to data, operations, Societal implications Stakeholders across the ecosystem will need to be ready to address a variety of societal implications driven by the increased adoption of AI across the industry The following three sections will explore each of these topics in detail Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 56 The New Physics of Financial Services |

58 AI opportunities a cross f inancial s ervices

59 Cross - sector opportunities | Overview institutions in a with the opportunity to change their businesses AI presents multitude of ways ranging from incremental improvement to complete reinvention trategies - driven c ross - sector s Overview of AI to improve existing processes to bold bets on new efforts AI investments by financial institutions are being made across a broad spectrum, from relatively conservative capabilities and business models What people mean when they say AI Pattern detection Foresight Customization Decision - making Interaction Doing the same things, better Doing something radically different dvice Leaner, d Tailored products and a making - faster Smarter ecision Ubiquitous presence o New value p ropositions perations Using advanced data science − Making products and services − automation to improve the Personalizing interactions to − − Differentiating offerings through − Using to more usual closely meet the unique available to customers in their as - efficiency of business new operating models and ways needs of optimize business outcomes - of working preferred format and channel customers rates) . lower default (e.g processes Expanding the reach of Reducing the − Building brand new products, − Integrating large volumes of data cost of simple, − Providing convenient, − − - quality service, while routine processes, while to derive better insights across services and business models high institutions’ channels and that use offerings geographically and core AI at the maintaining or improving quality units (e.g. better capital business maintaining scalability of experience allocation) across customer segments , highlighting The following slides will explore each strategy examples across industries and examining their implications Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 58 The New Physics of Financial Services |

60 o perations o Cross - sector pportunities | Leaner, faster expanding ever AI allows an set of processes to be streamlined, improving - and freeing up capacity costs efficiency, decreasing Why it matters e customer Suboptimal xperiences dampened r eturns Historically Increased regulatory b urden - Due to the presence of labour intensive processes that are Sustained periods of low interest rates, combined with pressure Expanded have increased operating regulatory requirements disconnected across the value chain, customers are provided . aggregators), have dampened returns, from new entrants (e.g costs, and will likely continue to do so as new regulations come e.g. slow payments) with a suboptimal experience ( - line growth increasing the need for cost cutting to drive bottom into effect (e.g . GDPR) faster eaner , perations Examples of l o Insurance management lending Deposits and Investment Payments time” in - “just Improve underwriting, pricing Offer seamless account setup - Provide Automate compliance and reporting and customer acquisition efficiency and accuracy lending AI can increase the efficiency of AI allows for the truncation of routine tasks, AI can contextualize data from multiple AI allows for analysis and decisions institutions reducing the friction across and underwriting by reducing error rates, to automate portfolio construction, systems (e.g . credit adjudication) to be made instantaneously, allowing credit to be increasing the speed of payments and automating datasets incorporating new speeding up key components of customer offered in processing and compliance reporting risk modelling onboarding real time Example Example Example Example - Intelligent data cataloguing Automated personal and AI ID verification through powered image recognition for compliance risk modelling business loans 120 109 98 Detail on Slide Detail on Slide 91 Detail on Slide Detail on Slide Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 59 The New Physics of Financial Services |

61 | Leaner, - sector o pportunities Cross faster o perations Streamlined operations alone will not be sufficient for most institutions, requiring these techniques to be combined with other AI applications o Key beneficiaries of l eaner , faster include perations Institutions offering , margin - Institutions with a high volume of Institutions facing high compliance low low per - requirements value account highly commoditized products accounts and short and medium term AI can reduce the cost of customer In the , Institutions with sizeable middle and - - back office operations in relation to their onboarding as well as customer providers of commoditized products overall size (e.g e.g . car insurance ) can use AI to service, allowing institutions with a large providers, ( . payment ) stand volume of customers (e.g develop competitive advantages in a ) retail banks . trading desks to benefit greatly improve service quality and expand from leaner operations, to market characterized by thin margins allowing them on costs to and cost - based competition refocus operational expenses reach, while controlling core competitive strengths Implications faster and Short - term cost cutting may spark a Customers will experience The role of human labour will need price as AI takes on a greater While . elevated levels of service as AI will to change “race to the bottom” on - processes, wide enterprise institutions can improve their operating requiring share of work, digitization of increase the efficiency and profit margins in the short service creating more intuitive and self - change management to transform processes, alongside human talent models alongside technology, as interaction term, competitors can easily replicate - improvements, making the long term when and where needed well as a societal response to workforce strategy uncertain sustainability of this disruption Sector Impact Sector Explorations Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 60 The New Physics of Financial Services |

62 advice and Cross - sector o pportunities | Tailored products off between cost and customization, allowing - AI resolves the traditional trade institutions to offer tailored products at near zero marginal cost - matters Why it isk r Heightened conduct ndustries i adjacent Influence of Increased commoditization Since the financial crisis, trust in traditional institutions has been Players in other industries (e.g large tech) are able to offer . Advice and personalization will become key differentiators as low, - increasing the need to offer high quality, and impartial, highly personalized experiences, leading customers to expect platforms and competitive forces drive products to become to grow and maintain customers’ confidence advice similar levels of tailored service from their financial institutions increasingly homogenized and a t ailored p roducts dvice Examples of Insurance management Investment Deposits and lending Payments highly with Drive loyalty by offering bespoke Provide detailed advice directly advisers Equip Increase the capabilities of sales real time to customers in personalized advisers incentives and rewards insights agents and analyse AI can be used to tailor rewards data across disparate the generation AI can of very AI can automate AI can be used to support complex programmes detailed, specific insights (e.g. personal making (e.g. quotes for - time sources, to individual customer - insights for real unlocking decision delivery to customers, creating a high sales - commercial clients), supplementing wealth report generation) to support behaviour, helping to maximize customer serve experience - quality, self advisers teams’ capabilities ’ interactions with their engagement clients Example Example Example Example Automated rewards dvice - AI financial a support for for analysis ad hoc Simple, Chatbot ustomers sales agents programme manager for c advisers retail 90 Detail on Slide 100 111 Detail on Slide Detail on Slide 117 Detail on Slide Implications Key Findings - Card Scenarios Opportunities Challenges Wild Cross - Sector Impact Sector Explorations 61 The New Physics of Financial Services |

63 sector advice products | Tailored Cross - pportunities o and Institutions with large and varied customer bases will likely be the primary beneficiaries of an increased capacity to tailor products at scale p roducts and advice include Key beneficiaries of t ailored Institutions with large and Institutions that serve a large diverse Institutions offering complex product customers volume of shelves financial products and services High offerings (e.g. barriers - net AI is - worth individuals have effective in breaking down Complex financial between sets mortgages) generally of data, creating an historically had access to highly require a higher degree of human involvement; personalized products and services. AI AI has opportunity for institutions with a broad suite of offerings (e.g . insurers, full the potential to reduce the burden on allows for similar customization to be service banks ) to offered at a significantly lower cost, human capital, allowing talent to focus unlock insights across products on higher making such products and advice activities (e.g. building value - and business units, improving the quality of advice and creating available to mass market customers personal relationships with clients) to service of quality improve the opportunities to deepen relationships retail . (e.g banking clients) Implications Third - party data will become a critical Customers will be offered more Advisory will be a critical competency , allowing resource , and the ability to set up and relevant advice and products customer loyalty in securing , helping institutions to set up virtuous cycles of individuals achieve their financial with financial and partnerships manage data will be a core area data that deepen relationships with their objectives and improve their financial non - financial stakeholders as a institutions seeking to customers by offering continuously of expertise for power well like - private banking - being experience is extended to mass - improving advice market their tailored advisory and product offerings consumers Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 62 The New Physics of Financial Services |

64 Cross presence | Ubiquitous pportunities - sector o by enabling better self reach - serve AI can help institutions expand their digitally more services to be delivered applications that allow Why it matters arkets m ew n in Traction xperience e Disconnect from underlying Shifting channel p references engaging new There is significant value that can be unlocked by Financial services are often a means to an end; a ubiquitous As customers’ expectations change over time, institutions need segments of customers within untapped markets, across presence allows institutions to tie directly with customers’ to ensure they are able to engage customers through their through digital channels countries, demographics and ongoing experiences without distracting from their end goals preferred modes of interaction (e.g . digital direct channels) p biquitous u Examples of resence lending Deposits and Insurance Payments management Investment first but unused Compete to become a provider of Use existing on personalized - Offer always Develop modern, mobile - insurance offerings invisible payments infrastructure platforms for distribution experiences across channels By integrating experiences into existing Institutions can expand their suite of virtual on AI can use a variety of data sources (e.g . always AI can be used to power - digital agents that interact with customers activities images, location data, sensor data) to party interfaces . (e.g third ), - offerings and capture new market share by enable real - time provisioning of insurance institutions can improve their reach and autonomously, providing support for using AI to offer a seamless experience that level of service to customers policies and instant claims handling basic questions and simple tasks automates the purchasing process Example Example Example Example Cashier Wealth platform integrated less - Insurance app using Conversational - and - AI agents brick Alipay into image recognition mortar store Detail on Slide Detail on Slide 100 120 Detail on Slide Detail on Slide 90 111 Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 63 The New Physics of Financial Services |

65 pportunities presence | Ubiquitous o sector - Cross AI allows institutions to support customers when and where they make decisions, but may also increase competition for customer engagement u include resence p Key beneficiaries of biquitous I where volume is a key nstitutions I whose products are nstitutions I serve - offering highly self nstitutions products and services used to execute transactions driver of profitability Physical cards (e.g In product categories where customer . credit, debit) are a Companies seeking to establish - life . ETF, economies of scale (e.g engagement is self means to an end (i.e . digital . directed (e.g acquiring a good banks), ensuring that customers are or service); integrating directly into insurance and mortgage providers) can use AI to expand the reach of their channels (e.g. able to manage their finances through party - third retailers) will their preferred channel will be critical to enabling them to models, distribution allow new opportunities the become to access customer and assets card of choice, maximizing the share of maximizing customer satisfaction and customers’ spend traditionally beyond their reach retention Implications more difficult Customers will have more Managing relationships with third Customers become adapt to Regulators will need to to ensure that will be critical as financial parties their seamless retain interactions increasing channel complexity and experiences with to lines to - blurring of industry party financial product manufacturers finances, with decisions often being increasingly occur in third - continue interfaces, diminishing the brand continue to have direct access to end entirely integrated into third to effectively govern party experiences (e.g . e - institutions and maintain the safety commerce users of products ) presence of financial institutions and stability of the financial system Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 64 The New Physics of Financial Services |

66 pportunities Cross - sector o | Smarter decision - making - AI can give making capabilities, unlocking institutions enhanced decision novel insights that drive improved performance Why it matters anagement m isk r ccurate a more Need for The comparison of ost c declining Saturation of traditional investment strategies The mounting risk - management cost of compliance and An ability to provide differentiated products and performance is and to outperform the market It is becoming increasingly difficult customized activities creates the need to develop increasingly increasingly important as customers and investors defer a identify high - potential investment opportunities through estimation and more accurate risk profiles and allow earlier risk broader set of decisions to algorithms common investment strategies Examples of s marter decision - making : Deposits and Investment management markets Capital lending Market infrastructure Find new and unique correlations Improve deal identification, Predict defaults with greater Improve trade speed and price and sales activities accuracy using dynamic execution methods pairing between datasets Identify unexplored patterns to the impact of price track Mitigate Use large volumes of data to map, movements and Inaccurate adjudication leads to higher outperform markets where traditional active - investing and losses on defaults and missed revenue companies to create market analyse execute at the best price by using AI to themes and identify buy/sell opportunities strategies are less attractive for investors optimize trade=execution who were from creditworthy individuals strategies erroneously denied loans Example Example Example Example of trade AI prediction Macroeconomic trends identification - powered Machine intelligence AI credit modelling learning price impact and cost for M&A analysis through machine 121 Detail on Slide Detail on Slide 128 Detail on Slide Detail on Slide 91 138 Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 65 The New Physics of Financial Services |

67 Cross making decision | Smarter pportunities o sector - - Smarter can deliver new value through improved returns making - decision and more resilient product performance marter decision - making include Key beneficiaries of s facing high risk of human Institutions Institutions with commoditized Institutions that compete on error offerings performance AI and automation can help reduce By using AI to make more abstract For institutions that have increasingly human error (e.g. in problem detection, connections and test complex models, commoditized products (e.g . lenders, institutions (e.g diagnosis, planning, execution) in sectors . investment managers) insurers), AI can be used to drive better decision - can identify where the actions of a few individuals making and ultimately create and execute new outperformance competitive advantages (e.g. lower and investment have an outsized impact (e.g. opportunities managers, market participants, trading price, higher returns ) distribute improved returns to customers venues), helping prevent losses Implications Talent strategies will become a Institutions will be able to more Both retail and commercial customers will benefit as source of competitive advantage as better, AI - effectively mitigate risks as - decision algorithms improve the financial institutions fight to attract and driven, decisions improve financial making - accuracy of outcomes their adjudication and retain highly sought after AI talent underwriting activities Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 66 The New Physics of Financial Services |

68 p o pportunities | New value sector ropositions Cross - AI allows institutions to redefine their core offerings, unlocking untapped segments and revenue opportunities through new products and services matters Why it boundaries New innovation of differentiation of ources s new Value from c new ompetitors Entrance of AI based solutions break traditional business constraints on - Institutions that can distinguish their product offerings beyond The risk of entry by non - traditional competitors such as large scalability and efficiency, allowing institutions to develop brand simple price competition will better position themselves for a technology firms is pushing institutions to bring new and new offerings that were not previously possible future dominated by platforms and algorithms innovative offerings to market to stay competitive ropositions p alue v ew n Examples of Insurance Capital markets management Investment infrastructure Market and Develop unique strategies and proxy data to insure new risk time pre - Develop real post - Deploy new order types to protect Use - new risk trade categories - management solutions investors from risks proactively investment products Introduce insurance for new risk categories products that anticipate and avoid Present profitability calculations for overall Offer Expand product shelves to address new cybersecurity, product portfolio positions by factoring in the cost of specific adverse impacts from predatory strategies, investment strategies or unique risk factors . (e.g - price quote insurance) by moving past the exclusive risk capital different that were previously under - served trade and the of instability and illegal activities impact use of historical data to price policies scenarios Example Example Example Example trade impact analysis - generation Pre - Next New order types Quantify cyber - risk based on AI smart - using AI ETFs using AI beta Detail on Slide 101 Detail on Slide 129 138 on Slide Detail 119 Detail on Slide Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 67 The New Physics of Financial Services |

69 pportunities ropositions Cross p value | New - o sector By leveraging AI to drive business model innovation, institutions can better contend with non - traditional competitors and commoditized markets alue p ropositions include Key beneficiaries of n ew v Institutions that lack proprietary clients Firms serving institutional Institutions offering simple, data ownership of customer products commoditized Firms underpinning capital markets, clients and offering AI creates an opportunity for providers As open banking increases serving institutional the of commoditized products (e.g . retail complex services (e.g . investment portability of financial data and the operate in areas with managers banks, P&C insurance ) to ) often build ability of new entrants to compete in financial services (e.g . imperfect products and unmet customer that deliver new value to offerings fintechs and incumbent firms (e.g . customers, allowing institutions to break large techs needs. AI can accelerate product ), innovation, allowing these institutions to ) must look to out of commoditized markets and offer online payment providers differentiated products and services at develop new value propositions to make major, transformational shifts in eroding data advantage how these services are offered a premium combat their Implications Underserved customers will have an Regulators will need to monitor Incumbents may have to shift to ensure innovation closely as more strategies to capture opportunities improved experience in and – traditional AI as competitors introduce new – non traditional institutions remain compliant throughout - the race to deliver more value through products products and services that may be and services become to or new services for available substitutes complementary existing offerings Sector Impact Sector Explorations Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 68 The New Physics of Financial Services |

70 Cross - sector c hallenges

71 sector Overview Cross - | c hallenges While the potential benefits of AI to financial institutions are immense, execution timeline to realizing value are often underestimated challenges and the In order to unlock the full value of AI, financial institutions will need to make significant changes to ready their organizations for the deployment of these new technologies Our research has identified the following as the most significant challenges to the implementation of AI in financial institutions quality data Institutions struggle to make available the large quantities of high - Data required to successfully train AI across their owned and unowned datasets Many valuable applications of AI require complex, deep and broad - reaching Operations on” implementations integration into the business, not just simple “bolt - AI fundamentally redefines the role of talent in financial institutions, and often Talent requires human capital to change at a pace that exceeds any past transformation outdated model Current regulatory frameworks were built based on an increasingly of the financial ecosystem, creating significant uncertainties for institutions seeking Regulation cutting to employ - edge implementations of AI and highlight examples of key challenges across industries and examine The following slides will explore each area of investment, their implications Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 70 The New Physics of Financial Services |

72 sector c hallenges | Data - Cross Incumbent institutions possess extensive stocks of data, but to often struggle deploy it effectively in AI applications learnin g - of Data is the single most important ingredient AI; the predictive capabilities of AI models are defined by the breadth, depth and quality of input data, as the machine algorithms) themselves are relatively easy to access methods (e.g . c ata hallenges d Notable Persistent - Fragmented internal d ata quality i ssues s torage data Insufficient data of depth and breadth “The value of AI will be seen when traditional and “Data is everywhere, but nowhere at the same time” “We must first deal with the extent of irregularities traditional financial data are combined” created by human error” - services non More mature AI applications depend on insights that span across a patchwork of Incumbents’ financial data is stored across Incumbent institutions already own large sets of financial data, but datasets, necessitating a broader diversity of data than financial specific disparate systems (e.g . the proliferation of product - much of it is not consistently formatted across the organization institutions have access data to. To navigate this complexity, new requiring systems) that ingest and display data in different ways, and may contain errors; this makes the data difficult for AI and partnerships established across traditional need to be - data significant engineering work to create large, and costly the cost of implementing new AI applications and increases - non traditional stakeholders common data lakes that can serve as inputs to AI use - enterprise capabilities across cases d partnerships of igitalization and APIs customers Inadequate understanding Incomplete Lack of data - “It is about time that paper “The future of financial services requires seamless “We need a holistic view of customers’ data if we are based and batch AI to optimize advice” access to data from across the financial ecosystem” to use processes were eliminated” Many institutions remain reliant - - intake formats time data flows across on analogue data AI is driving the need to build real must be able to access customer information across Advisers (e.g data, yet the lack paper) that require effort to be made institutions to access essential of . machine - . readable institutions and accounts to build a complete understanding of sharing is a roadblock Digitization of these inputs will improve data governance and commonly agreed standards for data - to customer needs. Currently, this data sits in various incompatible implementation until new partnerships and integrations increase speed, and consistency of processes accuracy prevents data analysis driven by AI systems and formats, which formed are from reaching its full potential Sector Impact Sector Explorations Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 71 The New Physics of Financial Services |

73 Sector Cross - challenges | Data Financial institutions must if customer data they are to make better use of continue to own the customer experience - specific implications Stakeholder Fintechs by finding excludable datasets carve out defensible positions can entrants (e.g., independent data providers), and other new , The increase in data the adoption of accompanies flows that Fintechs → AI will make many datasets increasingly commoditized and readily available. their niche technological expertise to develop innovative and differentiated datasets that complement financial data, can take advantage of positioning themselves as critical members of the financial services ecosystem Regulators will need to cope with the increasing complexity of data management, consider their role in protecting consumers w hil e facilitating innovation - sharing As → becomes more interconnected, the monitoring and auditing process for regulators will also become more complicated. The data improper use and increasing breadth and depth of data flows within and across organizations (e.g. through screen scrapers) increases the risk of data breaches, potentially leading to an increased need for regulatory oversight in data management and sharing Large technology firms have diverse data and leading - edge infrastructure. As they continue to search for complementary financial data, they stand poised to seize the initiative in building AI - enabled financial service capabilities as → technology companies such as Google, Facebook and Alibaba have positioned their enterprise strategies to acquire and use Large much data as possible. Given new regulations that make certain financial data more accessible (e.g . PSD2 ), these firms could move swiftly to obtain an expanded set of financial data and apply their technology advantage to inform the development of new capabilities Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 72 The New Physics of Financial Services |

74 Cross | Operations hallenges c sector - Legacy technology infrastructure and rigid operating models are additional hurdles to deploying AI within incumbent financial institutions ions remain reluctant to prioritize core technology AI capabilities must be tightly integrated with core systems infrastructure in order to drive value, but many incumbent insti tut - pay not have immediate do enhancements that offs hallenges perations c Notable o Extent of change to resistant process Intermediaries e - Legacy systems and technical d ebt r engineering ‘good enough’ aid solutions will no longer be - “Band “The way things get done in financial services is built to be incentives for influential parties need “There ” enabled with strong networks and domain knowledge to form to remain competitive in a world powered by AI processes will look around people. AI - different” radically a new ecosystem to accommodate AI” Transformation of processes across value chains requires Existing process maps in financial services are based on the Legacy systems operated by incumbent institutions have accrued from , who are themselves cooperation institutional intermediaries movement of information between people. As AI takes hold, new require significant technical debt and enormous overhauls to be put at risk of disintermediation by technology. very this These steps and structures process maps need to take into account new configuring APIs and adapting to . ready for AI implementation (e.g attempt to direct how may intermediaries resist change and support that machines and humans the interaction between real - time data flows ) operations might change based a rchitecture narrow Lack of agile , cloud Organizational structure r AI is and task - specific edesign - “AI must be retrained for each small step it needs only by “Institutions will require a clear view of the structure be unlocked “The full potential of AI can they need to capitalize on the benefits of AI” building support platforms simultaneously and to take, making end - end transformation labour - to - planning for change” intensive” AI will render aspects of the organization obsolete, requiring need To use AI at its highest potential, institutions to efficiently cannot yet be applied generally, and thus models must be AI . and enable updates with power store data, access processing organizational structures to become more adaptable (e.g leaner customized for each discrete - use case . This increases the need positioning of talent to ) as well as change the and more agile ease best enabled through core infrastructure built on this is – for both technical experts and domain specialists, particularly as structures agile, cloud adapt to new ways of working - based microservice are highly complex and processes services many financial obscure Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 73 The New Physics of Financial Services |

75 hallenges | Operations c Cross - sector Prioritizing technology and operations investments that will not yield an value in AI - term longer immediate return is essential to unlocking Stakeholder - specific implications must in costly and unglamorous core systems remediation to capitalize Incumbent financial institutions on the full benefits of AI invest → can be realized. Although these upgrades seem daunting, incumbents must consider these investments before the transformational benefits of AI m Incumbents can consider wholesale replacements of core infrastructure or phoenix - odels (e.g. establishing a challenger bank), as well as future proofing new infrastructure by building around microservice architectures Successful implementation of AI will require purpose - built executions of AI capabilities → AI trained to perform in a certain context will not be effective or applicable to other similar environments; successful depl oym ent of AI capabilities effort needed to will require significant forethought to ensure they can derive meaningful insight. As a result, incumbents are likely to misjudge the complete projects and f irms have a head start in AI capability development Fintechs large technology Both small and large technology players can take advantage of their technical advantage to enter the financial services market with a strong value → . On the other hand, incumbents may find it more difficult to compete, especially given that they must first invest in upgrades to core proposition infrastructure systems Simplification of market infrastructure will require a collaborative effort to ensure - in stakeholder buy → Changes to current market infrastructure must be made to allow for effective integration of AI capabilities into financial in sti tutions’ operations. F or example, data flows must be dynamic, uniform and easy to share across institutions. Collaboration between regulators and mark et participants is required to create a common model to be adopted across institutions so that the necessary developments can be undertaken Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 74 The New Physics of Financial Services |

76 sector c hallenges | Talent - Cross There is no playbook for how financial institutions should manage the talent transformation that AI will in the industry precipitate changing needs of talent. There is significant uncertainty as to how organizations respond to the Creating an AI advantage will require a fundamental shift in how institutions journey, and a significant risk ways practices and cultures stand in the way of creating effective new should of working steward talent through the AI that traditional policies, c hallenges i nclude alent Notable t and capabilities Uncertain Mismatch of Inflexible employment r elationships skills vision for change “AI transformation will require a new level of buy “Leaders are currently crippled by the talent they do “The traditional constraints of employment no longer in - and commitment to see change through” and do not have” serve the aspirations of institutions and workers” AI demands a scale of change across financial institutions that is Not only is there a severe lack of AI talent within the market, this Extensive effort is required to construct a new employment difficult to define and will be even harder to maintain. Financial structures that are inefficient in internal scenario is worsened by contract: one that favours flexibility over rigidity, rewards institutions do not appear to have a vision of how work will retain or transition the needs for talent their ability to recruit, productivity with protection, and structures skills and capabilities change, nor a - roll how to clear vision of out change across . deep domain across roles (i.e expertise and complementary rather than functions or tasks around solutions roles and teams individual functions, attributes to machines) eaders training c Narrow capacity Incompatible corporate ultures for Ill - equipped p eople l “It can’t be seen, but you can feel it. Culture is a get you...anywhere” here won’t “What got you “We know we need to train, but for what purpose, factor that builds you up or breaks you down” and to what end?” - there is an immediate need to “AI talent”, based environments, predicated on narrow scopes of Due to limited supply of Rules Internal fiefdoms need to be broken down and replaced with authority and routine work arrangements, have difficulty in invest resources in more proactive forms of enterprise - wide fluid channels for by characterized strategies for engagement, transitioning to a future that demands agility to be a core upskilling and awareness. These efforts are helpful in driving communication and staunch standards for collaboration. The change, but inhibited by the shortage of ‘trainers’, competency. While culture is nebulous, it serves as an essential pace of learning current capabilities of people managers to lead change are and natural limitations of individuals’ capabilities obstacle to effective transformation and are putting at risk the performance of inadequate, teams Sector Impact Sector Explorations Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 75 The New Physics of Financial Services |

77 sector | Talent Cross - hallenges c in recruiting and retaining people with the Financial institutions often lag knowledge, skills and capabilities needed to create an AI enabled workplace - Stakeholder specific - implications capabilities through strategic relationships with incumbents and/or demand” - “in Fintechs have the opportunity to become suppliers of regulators, as these institutions attempt to dramatically “upskill” their workforce for a new world of work talent has not been a priority for incumbent institutions, but they must adapt to be able to adopt AI skill sets . → Traditionally, high - skill technology e th lead now have an opportunity to form service partnerships across the financial services ecosystem to help institutions manage Fintechs that bank”, “changing the “running while maximizing benefits and limiting pain competing priorities of the bank” and technology are taking aggressive measures to secure their talent positioning Large through generous compensation packages and f irms strategies that target non - traditional roles With a limited talent pool that is fluent “war for talent” as institutions across industries try to gain a competitive edge. Firms with the → in AI, there is a compensation gid best compensation and culture will attract the best talent. Incumbent financial institutions have been slow to adapt their ri while large firms have moved rapidly to attract top talent into entry - level roles structures, technology Regulators face a larger talent gap than private institutions, challenging their ability to take confident positions on the u se of AI - → order to shape future regulations and conduct their macro - prudential duties, regulators must acquire an in In depth knowledge of AI and automation to correctly gauge capabilities and predict areas of risk and impact. Slow movement in upskilling regulatory workforces will create a chilling effect for firms seeking to drive AI applications that are both innovative and compliant Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 76 The New Physics of Financial Services |

78 | Regulation Cross - sector c hallenges to keep pace with emerging often struggle Existing regulatory regimes in technologies, creating roadblocks the deployment of AI capabilities Prescriptive regulations are currently limiting the advancement of AI in financial services. Inflexible requirements and regu lat ors’ limited resources constrain their ability to the rapid pace of change, creating significant regulatory uncertainty for institutions seeking to use new technologies keep up with c hallenges i nclude egulation Notable r m for identity u An odel Complexity of r ndefined egulatory f Lack of s rameworks tandards iability l “If AI powered decisions result in amplified losses, - “Increased flows of data are crucial to the “Institutions must consider how to develop AI while when who becomes liable?” enablement of AI, but are not without their costs” staying compliant, which is not the easiest path implementing new offerings” AI introduces ambiguity into responsibility in the event of loss or Realizing the benefits of AI requires of the continuous flow based solutions is - Interpreting regulatory requirements for AI - Institutions are hesitant to use third outcomes. negative party personal data across institutions. Robust authentication and complex, as these frameworks were not designed with AI services for AI systems as regulators will the primary likely hold that are standardized across institutions are verification methods applications in mind; this exposes institutions to making mistakes institutions responsible if there are damages to recover the risk of necessary to ensure security and privacy, and minimize in the adoption of AI harms fraud or other undue s sharing regulations ystems Fragmented data Tendency to avoid r isk - Auditability of new - Algorithm firms’ global ambitions “ “The success or failure of driven “Uncategorized risks introduced by AI into the AI solutions are often complex will black boxes even to the creator” be determined by the regulatory environment of their ecosystem are intimidating to tackle” local geographies” The fragmented nature of regulation across nations segments While regulators require audits of new processes (e.g. to verify the Regulators are continuously challenged to create the optimal independence), many AI processes are not easy to audit pace of AI innovation and the structure of financial services by balance between enabling innovation and mitigating risks. region. (e.g. traditional interim working steps are often skipped entirely), Strategic national competition also makes regulatory H owever, the complexity surrounding AI, combined with raising questions around when an audit is or isn’t necessary barriers for adoption and coordination difficult, creating stalling regulators’ natural tendency to be cautious, suggests that investment regulatory frameworks are likely to lean in favour of risk mitigation Sector Impact Sector Explorations Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 77 The New Physics of Financial Services |

79 | Regulation Cross - sector c hallenges Unlocking the full potential of AI will require financial institutions and their create new approaches and solutions - regulators to co Stakeholder - specific implications run into data - privacy , As institutions invest in AI, they will and liability concerns; the fear of mis - steps will slow down the sharing implementation of AI capabilities → Complex and fragmented regulations pertaining to data - sharing and privacy make it difficult for incumbents to pursue AI projects with confidence. Falling foul of emerging regulations comes at the cost of punitive measures and/or loss of customer trust. Regulators face increasing pressure to consumers, yet still leave room quickly understand developing trends and develop stringent policies that protect for innovation While the principles underlying consumer financial protection and fairness continue to remain sound, they will need to be reinterpreted for a financial system built with AI at its core apply to emerging AI technologies. However, the interpretation of The principles on which financial regulation was established remain relevant and → hile adopting new these principles is not clearly defined, and regulators must adapt “black - letter” regulations to remove redundant requirements w rules to protect against new risks Diverging regional data sharing regulations make international coordination on macro - - prudential oversight harder regions. This challenges institutions that operate globally and necessitates and regulation are apparent across data standards Differences in → an impact across jurisdictions. Risks such as rogue trading, capital international frameworks to manage common systemic issues that would have - cybersecurity and adequacy border collaboration to reach a resolution require cross Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 78 The New Physics of Financial Services |

80 Societal implications

81 Societal implications | Overview it will create unintended consequences for As AI transforms the financial system, society at large that require public - private cooperation to address fully of AI in Emerging societal issues magnified by the use ervices financial s risk disruption Ethics and discrimination Workforce Systemic created a fear of The onset of AI has increasingly critical an As AI takes The enigmatic nature of AI scale labour displacement; day operations of - to - role in the day large technology may seem like “magic” - to strategies must be developed to the financial system, outsiders, but understanding its it poses a new source of systemic risk that has the critical to detecting and is behaviour effectively manage the forthcoming models preventing talent shift and transition large potential to disrupt national and that discriminate exclude marginalized global economies, necessitating new against or portions of the workforce through the evolution groups and individuals R ndustrial I Fourth controls and responses industry ces The following slides will explore the broader challenges faced by society as a result of AI innovation in the financial servi Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 80 The New Physics of Financial Services |

82 implications | Workforce disruption Societal The effects of AI on the labour market are a significant concern globally, especially the implications for those in and middle - complexity roles low - – rather, understanding Discussions of competing utopian and dystopian narratives. Neither of these visions is accurate impact on the financial workforce often feature AI’s changing the true impact on labour requires a deeper examination of how the financial ecosystem is volution Predicted labour m arket e Future scenario cenario s Current Supply of current capabilities Complexity of workforce transformation and different experiences of talent Displacement in demand for some A traditional capabilities Supply of new roles - term because of a near Roles being replaced emphasis on operational efficiencies Demand for new skills but uncertainty about the replacing old modes of skills Future purpose and positioning of roles required in the future Characterized by new roles, B operating and ways of working structures Within t raditional institutions level opportunities Displacement by a low reduction in - Within supporting industries Growth of roles in ne w members of the C financial ecosystem New roles in supporting industries demand for some traditional Growth of roles in new members of the Displacement in old modes of replacing skills Future B A C capabilities operating financial ecosystem cutting strategies, the Institutions need to source more technical skills to develop As institutions pursue efficiency and cost ecosystem party providers outside - role of third The - - , new entrants) that complexity roles are likely to be and middle - routine low intechs firms, f abilities to value technology - (e.g . large AI solutions, as well as more high displaced (e.g. back office). Since these roles constitute a complement them ( e.g. creativity , ingenuity, insight). In supply services to financial institutions will grow, and these order to unlock the full potential of this workforce, roles are likely to look radically different. They will require significant portion of jobs in financial services, there is a institutions must effectively re skill sets (e.g. - task existing talent as well as technical backgrounds), have different new risk of a net decline in overall roles available, especially in models the nearer term source new external talent and have different compensation cultures Sector Impact Sector Explorations Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 81 The New Physics of Financial Services |

83 Societal i mplications | Ethics and d iscrimination The use of AI in financial services introduces new ethical pitfalls and risks unintended bias, requiring the industry to reflect on the ethics of new models his demands proactive collaboration between institutions and regulators While the benefits of AI are clear, the potential unintended consequences are less easy to visualize. T to identify and address potential sources of bias in machine decisions and other exclusionary effects making methods, there is affect for discrimination that will test societies’ moral obligations if decisions adversely As AI introduces new disparate groups decision - potential Bias in input data discrimination towards training drift - Post Bias in development As AI systems self among development Subconscious bias or lack of diversity - improve and learn, they may Bias present in input data, as well as incomplete or acquire teams be unrepresentative datasets, will limit AI’s ability to have unintended consequences new behaviours that may influence how AI is trained, carrying bias objective forward Continuous Model Input learning Furthermore, AI may make it harder to explain solutions compounding the impact of potential discrimination by making safeguards more difficult to establish , se u Auditability nput i over Uncertainty models Unpredictability of As market conditions evolve, it may be difficult to predict Regulators often lack the technical expertise to inspect Some methods of AI training may obscure how data is creating the potential for discrimination algorithms, especially if development is improperly how models will respond, with the resulting portfolio and used in decisions, implications documented or there are persistent, system wide gaps in - macro (e.g . using race or gender data in credit decisions) governance not always a critical factor. For the most Despite these risks, most AI used today is not a “black box” and, depending on the use case , the ability to explain solutions is Note: complex AI applications, parallel models can also be run that either predict or create confidence in the behaviour of the AI model Sector Impact Sector Explorations Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 82 The New Physics of Financial Services |

84 | Systemic risk implications Societal Without proper oversight, AI innovation could introduce new systemic risks into the financial system and increase the threat of contagion part of the financial ecosystem, they will transform the landscape of existing systemic risks while potentially adding new, unexpected risks. As AI models become a pervasive include the potential for herd behaviour among AI systems and the increased vulnerability of financial operations to of particular concern Areas cyberattacks cyber - risk Herd risk and consolidation Increased Risks of contagion interconnectivity as AI demands increased digitization Increased , where losses from negative market events by a shared algorithm passing decisions across and linkages across institutions, creating more opportunities for may be increased cyberintrusion multiple institutions Systemically important technology and third - party data providers on Increased risk of errors as model miscalibration would impact on are critical in developing AI systems, increasing the reliance many institutions simultaneously (e.g . miscalculating credit risk) previously unregulated stakeholders No one institution, government or regulatory body can address and prepare for these challenges alone, necessitating increased en gagement and collaboration across all parties in the global financial system Sector Explorations Sector Impact Key Findings Wild - Card Scenarios Opportunities Challenges Implications Cross - 83 The New Physics of Financial Services |

85 Opportunities, challenges, societal i mplications | Conclusion AI is likely to have the global financial system – the a transformative effect on task of the ecosystem will be to maximize benefits while mitigating harms Regardless of sector, AI presents opportunities for financial institutions across their value chains: from Opportunities meaningful improvements to business - as - usual processes to radical, industry - changing plays environments, Incumbent institutions face significant challenges, across their operating to achieving Challenges the full value of AI. Overcoming these obstacles will require significant investment and discipline to - driven agendas deliver on the ambitions of AI As the global financial system undergoes transformation, institutions, regulators and policy officials Societal implications scale displacement of labour, as well as develop modern tools must be proactive to manage the large - to manage new ethical uncertainties created by automated decision - making - Card Scenarios Key Findings Cross - Sector Impact Sector Explorations Wild 84 The New Physics of Financial Services |

86 How is AI being deployed by sectors of financial services? Sector explorations

87 Sector e xplorations | Overview Our views on emerging competitive dynamics are informed by a detailed examination of AI enabled strategies across every sector of financial services - section narrative below, which Each sector will follow the four - detailed provides a Over the following slides, we will explore the following six sectors in detail overview of high - potential AI strategies 1. Sector overview - high A trends level overview of the sector, its emerging Insurance Deposits and lending and the key challenges that it faces 2. AI’s anticipated impact on the sector the impact AI is poised to have on the An overview of sector and an introduction to selected strategies that AI is poised to enable or accelerate Payments Investment management Selected strategies enabled and enhanced by AI 3. A detailed review of each high - potential AI strategy, including evidence of its importance as well as specific opportunities observed in the marketplace today Note : Examples range from automation techniques (which may not meet some readers’ bar for what constitutes “AI ”) through to advanced AI techniques. These examples are intended to highlight key industry and technological trends and should not necessarily be taken to represent the best or most advanced practices Capital markets Market infrastructure 4. Looking forward A summary of key implications for various stakeholders (e.g. individuals, financial institutions and regulators) Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 86 The New Physics of Financial Services |

88 Deposits and lending Sector exploration

89 Deposits and lending | Sector overview margins are under significant pressure, caused by increased Core banking regulatory burdens, accommodative monetary policy and new competitors I s s u e s f a c i n g t h e s e c t o r O v e r v i e w o f t h e s e c t o r cope s chapter and description Sector 48% expected annual growth rate of marketplace lending Increasing pressures on − 1 from traditional players from 2016‒2024 new competitors lending Deposits and of the retail and commercial banking sector. Institutions offerings are the core − Two out of every three PSD2 bank executives believe receive deposits (by offering customers deposits both secured finance and use these accounts) to will have a negative impact on profit margins because of loans (e.g cards). e.g. credit . mortgages) and unsecured loans ( 2 entrants new on this that facilitate these products are banking value chain; transactions This chapter focuses % of individuals globally Only 54 financial and Sustained low trust − “trust” covered in the Payments and Capital chapters. markets 3 consumer confidence rated sector in institutions, making it the lowest - 2018 scope Chapter (e.g. United States, European − Major global markets Sustained cost Financial Transactions Distribution Deposits Lending taking - Risk pressures across banking Union, China, Japan) are all expected to increase interest advisory 4 resulting operations rates to Q1 balance sheets 2019, in pressure on P2P • • Balance • Secured Branch • • Fixed • Savings - term B2B, C2B, B2C • Unsecured sheet • • • Phone Current Spending • maintaining existing − 75%‒ 80% of bank IT spend is on • Securitization Borrowing • Digital • core banking systems, and is expected to continue to Retail customers 5 profitability hinder sized customers Small and - b usiness medium globally expect the cost of 89% of industry executives − Corporate customers 6 compliance to continue increasing from 2017‒2019 Sector trends Almost 40% of adults globally do not have a bank Significant global − as global economic confidence continues to recover and Upward interest rate trajectory → 7 unbanked and account, or access to a financial institution central banks begin to tighten monetary policy underbanked populations have a 67% of bank regulators in 143 jurisdictions − (e.g Growth in low microfinancing and SME lending → as the cost of ) . principal financing - 8 mandate to promote financial inclusion servicing loans decreases, resulting in low - value credit products becoming more attractive Focus on customers as changing customer preferences for digital and personal advice results → in changing revenue models for depository and lending institutions − Regulators in the US and Europe have fined banks High costs of misconduct $342 billion since 2009 for misconduct, and this is likely → (particularly in mortgages) as regulatory and supervisory Shift in financing to capital markets 9 to top $400 billion by 2020 leveraging - de bodies continue to emphasize the of depository institutions Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 88 The New Physics of Financial Services |

90 Deposits and lending | Strategies summary AI can improve banks’ profitability through the delivery of personalized advice at transformation of lending operations scale and the deposits The rise of AI will initiate and accelerate the following changes in lending and : The range of potential customers The cost of funds in lending will The line between deposits and Lending products will become lending will become as deposits become more will increase as as are increase smaller and more agile as they blurred - traditional data makes it putting cash - flow management will be expensive to service, non customized to specific uses possible to serve new customers “as a service” pressure on net interest margins delivered regions and operate in new include: and l deposits enabled strategies in - Key AI ending Strategy C Strategy B Strategy A Offer automated working - capital Focus retail banking on improving Increase the efficiency and scale of solutions for commercial clients customer outcomes retail lending B C A treamline → E.g . u se machine learning to predict end → E.g . u se advanced analytics to s . - to - the end → E.g onboarding process to more efficiently cash flow events and proactively - understand a broader set of data to advise customers better adjudicate commercial lending their spending on evaluate and issue new loans and saving habits The following slides will explore each strategy, detail the components and highlight key examples Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 89 The New Physics of Financial Services |

91 b anking Deposits and lending | Strategy A: Focus r etail ustomer utcomes on i mproving c o A advice scale and at the moment of AI is allowing institutions to deliver at need, redefining experience banking the retail the value proposition of W h y i m p r o v e d c u s t o m e r o u t c o m e s m a t t e r savings c redibility d ifferentiated o isk fferings conduct Increasing New r o fferings t hreaten f ee r evenue p re is a entrants New threatening roviding 10 promising higher are New entrants with new products As banking institutions lose the trust of customers, The next generation of personal financial management - for on deposit interest apps are aligning their service and compensation with deposits and eliminating fees to loyalty will decline, making it harder to retain demand customers’ financial outcomes, threatening ownership of attract more volume levels and fend off new entrants the retail banking customer experience O b s e r v e d o p p o r t u n i t i e s experiences across channels advice in real time Provide personalized, practicable on Offer tailored, always - Pain p oint Pain p oint Often, financial advice is generic and delivered by generalist staff during limited Customers lack access to timely, clear and relevant financial information interaction points limited have and offline channels Current online ability to provide insights to customers on their Lack of clear and readily financial and the impact of those habits on their future financial needs. habits, Generalist staff (e.g . advisers ) are limited by the hours when they are branch employees, phone leads with customers. This limits the ability of these staff to customers to make suboptimal limited personal interaction accessible information through online and offline channels available and their market on personal advice to the mass - deliver always funds, resulting in worse financial health decisions on the allocation of apabilities New c apabilities New c nsights generated Intelligent employee dashboards can  Automated interactions provided by “chat   I Continuous  Process automation   A true single - customers to access and customer view generate insights from past customer bot” solutions allow can from improves connectivity third - party data of monitoring . on be achieved by interactions (e.g . analysis of calls ) social, mobile) accounts held across (e.g and and accuracy across advice through the - always receive recommendations to front - institutions line staff expanding network of digital channels provide different connect advisory to business functions intelligently scanning to day reality internal data held can - and accounts the day - raise proactive across silos alerts of retail customers AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples Clinc and other conversational agents allow financial institutions to automate Clarity Money Personetics, Fini by product “genius” Albert’s bills integrates account balances, compiles financial data them to provide personalized and responsive information and and credit scoring to help customers understand rote and common interactions, within and across institutions to suggest next enabling advice through digital channels best actions (e.g . switching utility providers to their complete financial story and actively effective options) manage their accounts more cost - Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 90 The New Physics of Financial Services |

92 lending scale Increase the efficiency and | Strategy B: of r etail l ending Deposits and B AI can deliver smarter and more nimble workflows that improve the productivity and reach of lending operations W h y e f f i c i e n c y a n d s c a l e m a t t e r time nclusion Expectations for real - Agile innovators are rewarded with higher profitability a ccess i improve Pressures to mature Institutions are expanding market share in Margin Connected and agile workflows, as well as the removal of - positions, even in high competition lending and markets, are the time additive customer non the for institutions that effectively improving interactions, are reducing emerging markets by ingesting data to better predict - 11 default, improving confidence in lending decisions it takes to access credit risk of reduce manual efforts in the back office O b s e r v e d o p p o r t u n i t i e s - Miniaturize unsecured lending to be use - specific Provide “just - in time” lending Predict defaults with greater accuracy Pain p Pain p oint oint Pain p oint based on the Inaccurate adjudication results in increased default costs and Loan origination can take days or weeks due to fragmented or priced Unsecured lending is not adjudicated use case missed revenue due to denials of creditworthy individuals intended purpose of an - data storage and manual entry and is dataset lines of credit) do not adapt to Data is stored across multiple internal and external systems, . Lending products (e.g credit . (e.g Adjudication is often based on a limited individual inaccuracies limiting the ratings), often manually entered. This increases the purchasing a new computer); number vs. . e.g purchases ( buying a coffee of reliability of creditworthiness calculations, allow for dynamic pricing based specific would such as non particularly for specific segments of the population - - and opaque , slow and creates delays in response time, resulting in making loans use unresponsive residents and unbanked or underbanked groups on the risk profile of a given transaction processes apabilities New c apabilities New c apabilities New c s ata analysis d party  Alternative data ources can - Third  Dynamic sourcing of data  language Natural  Advanced credit - decision  making - decision Automated  allows faster and (e.g can adjudicate customers in . be used in place of traditional SKU level data ) from disparate sources aid models can p rocessing that use machine how error free document reading to learning can improve the understanding of eliminates the need for manual - at low cost, assess credit scores to and real time segments allowing tailored lending data entry – opening the way ( e.g support activities such as . confidence of lenders to creditworthiness in transactions affect the extend credit, reducing creditworthiness of an scalable digital onboarding for for which data is not readily at terms) custom repayment information verification, user available identification individual and approvals the point of purchase and servicing channels defaults and expanding reach Examples AI and/or Advanced Automation AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples Upstart Affirm aims to streamline the provides businesses with financing solutions for their ZestFinance and monitors cleanses, Kyckr uses alternative Qudian makes personalized process by credit accessible online for origination - entire loan data and machine learning to remediates data to meet KYC customers, offering credit at the point of purchase; Affirm settles the automating activities from data increases consumers who need mobile full amount with businesses and then customers pay Affirm back build credit models, allowing obligations, and onboarding efficiency lenders to gain confidence in loans for access to small entry to verification tasks over time, with interest discretionary spending lending to new segments Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 91 The New Physics of Financial Services |

93 solutions for commercial clients | Strategy C: Offer automated working - capital lending Deposits and C AI is launching a commercial banking renaissance through improved data that unlock a vast underserved market integration and analytics tools W h y b e t t e r c o m m e r c i a l l e n d i n g m a t t e r s arkets commercial Bloated Threat of p rocesses m commercial lending Underfunded new e ntrants cut operations has cost of commercial lending The high Accounting and invoicing platforms are integrating lending Commercial lending for small and medium - sized 12 the profitability of incumbents and is pushing customers in into their business services platforms, encroaching on the businesses is under - served, particularly in emerging 13 for financing traditional market share of incumbent institutions capital markets markets some regions into O b s e r v e d o p p o r t u n i t i e s decision - making Automate and augment business credit Improve client advisory by integrating into data streams for opportunity discovery Pain p oint Pain p oint decision Incumbents have no real - time visibility with regard to the financial situation of their clients making - decision is significantly more costly than retail credit making - Business credit due to the need for manual analysis of business financial documents struggle to adopt sophisticated of Lenders credit models that incorporate detailed considerations liquidity into calculations, time the detailed, real to they lack access because - and Commercial lending products are costly to originate as applications must be reviewed manually to profitability , growth on their own to Financial information is reported in non - standard documents (e.g . income industry - and contextualized (e.g specific) data required. As a result, borrowers are left . determine creditworthiness. consuming process, especially - determine their own financing needs and navigate the lengthy, opaque process of commercial statements), making verification of these documents a manual and time lending SMEs for apabilities New c apabilities New c  Integration into client Enterprise Resource Natural language processing and  allow  Predictive algorithms can assess and Augmented analytics tools  dashboards , Planning (ERP) and data sources real time anticipate client borrowing needs in to access underwriters to customize their credit analysis can collate automated allowing time raw data, enabling lenders to get a advisers rapidly, using a variety of data sources to important statistics by deciphering complex to dynamically generate real - increase their certainty of - standard financial documents, offers and extend funds seamlessly deeper understanding of client fundamentals creditworthiness and non and offer more customized advisory services extracting relevant information in summary and ultimately credit adjudication form for underwriter review AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples Tradeshift and HSBC alternative to quickly originate bespoke loans for OakNorth’s “ACORN Machine” billion in analyses $3 has made over Amazon Lending are partnering on a data 14 two years small and medium - enabled by It has merchants globally, loans to small sized businesses. grown just in platform solution to extend working capital to its loan book to £800 million and 16 “as a service” analysing companies by viewing and based adjudication technology lenders to other cloud the entire - insight into the revenue flows of merchants that is offering its 15 - commerce platform supply chain of commercial are part of its e clients Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 92 The New Physics of Financial Services |

94 Deposits and lending | Conclusion Looking forward form the by algorithms As customer experiences are increasingly informed accounts may no longer deposit , 1 experience for customers centre of the banking to deliver digital advisory may make them the natural owners of he emerging ability of platform solutions T 2 customer relationships in retail banking engineered, the shape of teams and the - As roles in advisory and adjudication functions become re 3 composition of talent in these areas will be transformed decision - making models will make lending decisions more accurate, but will also raise ethical New 4 and decision opaqueness biases questions regarding potential decision will raise questions about the making - treatment of on data for financial A reliance personal 5 information identifiable Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 93 The New Physics of Financial Services |

95 Deposits and Lending | References References “P2P Lending Market In Numbers”, as of 29 September 2017. Getline https://medium.com/getline - in/p2p - lending - market - in - numbers - the - getline - network - 4299da9ad867 1. . Retrieved from - “European Deloitte . Retrieved from https :// www2.deloitte.com/content/dam/Deloitte/nl/Documents/financial - services/deloitte - nl Survey, as of October 2017”. fsi - psd2 - survey - 2. PSD2 - highlights.pdf results “Customer 3. rust : Without I t , You’re J ust A nother B ank”. EY . Retrieved from https ://www.ey.com/Publication/vwLUAssets/ey - trust - without - it - youre - just - another - bank/$ FILE/ey - trust - T without it - youre - - - another - bank.pdf just 4. “Interest Rate – Forecast – 2018 – 2020”. Trading Economics . Retrieved from https :// tradingeconomics.com/forecast/interest - rate is 5. egacy Is N ot E nough”. Deloitte . Retrieved from https :// www2.deloitte.com/content/dam/Deloitte/global/Documents/Financial - Services/gx - fsi - us - why - legacy - L - not - enough - “When 2008.pdf Year Ahead”. “Opinions Global F inancial S ervices Regulation and Industry Developments 6. on Duff&Phelps . Retrieved from https ://www.duffandphelps.com/ - for the / media/assets/pdfs/publications/compliance - and - regulatory - consulting/2017 - global - regulatory - outlook - viewpoint.ashx not 7. % of the World’s Population Does Not Have a Bank Account”, as of 21 November 2016. Gomedici . Retrieved from https ://gomedici.com/39 - of - the - worlds - population - does - “39 - have - a bank - account / - 8. “The Foundations of Financial Inclusion”. Worldbank . Retrieved from http :// documents.worldbank.org/curated/en/348241468329061640/pdf/wps6290.pdf Banks’ Misconduct 9. “U.S ., EU Fines on :// to Top $400 Billion by 2020: Report”, as of 27 September 2017. Reuters . Retrieved from https www.reuters.com/article/us - banks - regulator - fines/u - - eu - fines - on - banks - misconduct - to - top - 400 - billion - by - 2020 - report - idUSKCN1C210B s “ 2016 EY . Retrieved from http ://www.ey.com/gl/en/industries/financial - services/banking --- capital - markets/ey - global - consumer - banking - survey - 10. The Relevance Challenge”. “The Digital Disruptors: - Banking Got Agile”. Accenture . Retrieved from https :// www.accenture.com/ca - en/insight - outlook 11. digital - disruptors - how - banking - got - agile How “Small Businesses Becoming More Satisfied with Fintech Lenders: Survey”. 12. . Retrieved from https :// www.americanbanker.com/news/small - businesses - becoming - American Banker fintech - - with - satisfied - lenders - survey more 13. “Eying Emerging Markets, Lenders See A Wealth of Opportunity”, as of 2 March 2015. Pymnts . Retrieved from https ://www.pymnts.com/in - depth/2015/eying - emerging - markets - lenders - see - - wealth - of - opportunity / a “Amazon . Retrieved han $3 Billion to over 20,000 Small Businesses”, as of 8 June 2017. Business Wire 14. from Loans More t Billion :// https Loans - 3 - www.businesswire.com/news/home/20170608005415/en/Amazon - 20000 - Small - Businesses - 15. “HSBC and Tradeshift Join Forces to Revolutionise Working Capital Financing”, as of 30 March 2017. HSBC . Retrieved from https ://www.hsbc.com/news - and - insight/media - resources/media releases/2017/hsbc - tradeshift - announcement - final - 16. “Entrepreneur - Focused Bank Oaknorth Hits ‘Unicorn’ Status in Funding Deal”, as of 13 October 2017. Business Insider . Retrieved from http :// www.businessinsider.com/oaknorth - value - - the clermont - group - toscafund - coltrane - invest - 2017 - 10 Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 94 The New Physics of Financial Services |

96 Insurance Sector exploration

97 Insurance | Sector overview - particularly due to the driven transformation, Insurance is highly sensitive to AI rising influence of new entrants O v e r v i e w I s s u e s f a c i n g t h e s e c t o r s chapter and description Sector cope erception than banks, supermarkets, car − Poor consumer Insurers are less trusted p 2 of insurers and online shopping sites manufacturers is a means to protect the discretionary income of customers by spreading or insurance At its core, − reports outline negative practices that Consumer Insurance is externalizing risks. most attention from receives the both one of the sectors that of interest between the highlight perceived conflict and disruptors looking to challenge the traditional order of business. Thus far, traditional investors 3 consumers and insurance companies insurers and carriers have maintained their positions of strength due to the complexity of products and advantages of scale, which allow them to better pool and hedge risk. 57% − consumers, across all product insurance of global Customer expectation of types, prefer to hear from their providers at least increased nteractions semi - i Chapter scope 2 47% contact currently receive that level of annually; only General / P&C Retirement / Life insurance Reinsurance Distribution annuities insurance f purchased stakes in Ping An Alibaba and Tencent − Rising involvement rom 4 • Brokers Retail and • Covered in • • Life • life Whole echnology p l Insurance in a $4.7 billion deal in 2014 arge t layers • Digital life life • Investment • Term commercial Non - platforms Management • Insurance - • Motor joint venture with − Amazon announced a health insurance Comparison • linked Health • 5 Hathaway and Berkshire JPMorgan websites • securities Home Marine • 6 2018 XL Group for $15.3 billion, March acquired roperty in p Consolidation − AXA c i nsurance and asualty Sector Trends 7 2018 − acquired AIG Validus for $5.6 billion, January markets acquired − Allianz significant stakes in Euler Hermes and , Autonomous → vehicles change the automotive property and will casualty insurance landscape 8 LV’s general insurance business, 2018 February coverage responsibilities from drivers to automotive manufacturers likely shifting risk - disruptors “Insurtech” → have been more successful than fintechs in other sectors, particularly 1 - South growth of 3x – Growth potential 2x − life - non and life Asian East within the distribution space, where new business models and platforms are scaling rapidly 9 average compared to the global c insurance markets oncentrated in non - and health, are automotive of common personal lines, such as property, Commoditization → traditional arkets m − growth in gross written premiums in China between 80% pushing incumbents to look for volume in new types of insurance 9 2010 and 2015 → are becoming more complex to meet customer expectations for Distribution networks average real growth of the global insurance industry − 3.7% channel experiences - increasingly digital and omni 10 2018 from 2016 – Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 96 The New Physics of Financial Services |

98 Insurance | Strategies summary AI will and help insurers predict products risk with greater accuracy, customize use enhanced foresight to rapidly deploy new products The rise of AI will insurance changes in : initiate and accelerate the following will get more for their Customers Distribution will permeate devices more office underwriters - Brokers, back Insurance will be time and will be at risk of become as policies will money and geographies in order to cover and adjusters connected, more real - the situational needs of customers more accurate as new algorithms cheaper and coverage will be more as operations across disruption functions become streamlined and new data take hold complete Key AI - enabled strategies in insurance include: Strategy D Strategy A Strategy E Strategy C Strategy B Streamline operations ervices Improve and expand Insure against new types that to Develop a differentiated Offer add - on s win experience nsurance i complement claims of risk in new ways distribution strategies on price E D C B A se image recognition and se → E.g . u se alternative data and u → natural language → to time monitoring - E.g E.g . u se predictive analytics . E.g → . u → E.g . u se real behavioural - exposure pricing to advise clients on risk dynamic and new data sources to fraud detection to speed up processing and advanced to lower risk develop unique insurance claims processing improve sales efficiency and strategies decision trees to improve markets into new underwriting and capital expand risk products that cover new efficiency categories key examples highlight the components and detail The following slides will explore each strategy, Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 97 The New Physics of Financial Services |

99 to Insurance | Strategy A: Streamline operations win on price A insurers monitoring give to AI is driving efficiencies in underwriting and risk commoditized markets edge, particularly in a competitive W h y w i n n i n g o n p r i c e m a t t e r s ates c Low h eadwinds Rising interest ratios insurance P&C retail Commoditized r for commercial nsurance i reate The high cost and involved nature of commercial risk The rise of platforms and policy standardization has left Rising interest rates will depress bond portfolio yields, and requiring better capital and for retail products such as differentiator price as the only assessments are threatening the profitability of hurting P&C insurers 11 home commercial P&C i nsurers car and management i nsurance portfolio O b s e r v e d O p p o r t u n i t i e s and accuracy time risk monitoring - Improve underwriting, pricing efficiency Increase capital efficiency through better risk modelling and real oint Pain p Pain p oint high Back testing and model validation is a cost process whose outcome has significant Even in digitized systems, underwriter workflows often require manual review or data entry for - non impacts on balance sheets and overall insurer profitability standard cases and exceptions - nascent and evolving understanding of risk limits the in modern process as our Risk modelling is an imperfect or slightly irregular cases that are not underwriting engines, there are “complex” Even rote labour programmed explicitly - intensive . but requires a high level of The costs for productivity of data scientists, making this critical process highly . These require manual review, which is often on the part of increasingly insurers are compounded by domain - specific knowledge changing costly talent for modelling regulatory requirements New c New c apabilities apabilities  Non New and unstructured data can be used in - static and complex decision trees can character r ecognition (OCR) can  Automated modelling using machine -   Optical systems allows institutions to run read, verify and standardize supporting be built with machine learning by analysing learning risk analysis using machine learning, allowing past cases, allowing for automated for a wider variety of situational thousands or millions of trial models a day, documents, eliminating the need for manual simulations. underwriting and pricing of complex and reducing the cost of meeting regulatory This results in more review accurate levels of irregular cases requirements liquidity capital AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples Swiss Re is using IBM Watson to Concirrus Quest Marine analyses unstructured datasets using AI increased payout Cytora enables faster OmniScience Fukoku Mutual Life calculation efficiency by 30% using AI, breaking provides live marine weather, reducing – insurers model runs for new risk analyse the impact of broad to find patterns. This helps insurers to improve loss and expense ratios, and to identify insurers and port data to even on the cost of its investment in automation risk management and the cost of portfolio and its market trends on fleet 12 13 warning when a vessel profitable segments for commercial insurance years improving overall risk metrics in less than two better price risk (e.g . enters an exclusion zone) Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 98 The New Physics of Financial Services |

100 c Insurance | Strategy B: Develop a differentiated laims e xperience B evaluate that are more claims, creating workflows AI is being used to accurate and responsive to customer needs W h y c l a i m s e x p e r i e n c e m a t t e r s c poor Avoid xperiences entre c ost c e New entrants digitizing claims major laims is a Claims fraud 14 Incumbents can address reputational concerns by offering filed. Some estimates put fraud at 10% of all claims Platforms and direct players are developing innovative 15 differentiated claims - claims channels, including online and interactive digital handling stories that result in positive Given that the average profit margin is 6% (US), claims solutions, to gain market share lowering fraud is a major profit opportunity customer outcomes O b s e r v e d o p p o r t u n i t i e s Triage and grade claims to increase adjudicator efficiency Process claims instantly Reduce fraud using new tools and new data oint Pain p oint Pain p oint Pain p Fraud is a major avoidable cost for in a insurers Any wait time for claims adjudication results Claims processing often requires manual review of complex poor documents, which slows response time customer experience Since claim forms are self attested, there is high potential for fraud. - Responding to claims is a lengthy process that involves numerous Fraud losses shrink insurers’ profit Claims documents are long and time - consuming to read and and are sometimes margins underwriters to evaluate injury or steps and approvals. Often this means customers must pay out of passed on to customers in the form of higher overall premiums interpret, requiring highly skilled pocket and wait with uncertainty by a policy damage and determine whether about timelines and outcomes it is covered apabilities New c apabilities New c apabilities New c (e.g .  Analysing large quantities photographs, IoT  Rank claim severity  Provide adjudicators with using   Use of new data to verify damage Analytical models using (e.g . news summaries and statistics sensors, weather data) the moment a claim is filed allows efficiently using of data deep learning to read claims external data that enhance their decision reports, social media) can - machine learning allows for documents and score their institutions to extend initial funds that can immediately address making severity depth review - in urgency, the and can increase of every cases of more accurately flag while reducing the chance of fraud needs, customer and fraud of individual efficiency submitted claim , reducing losses while compliance to expedite triage fraud for underwriters increasing throughput AI and/or Advanced Automation Examples Examples AI and/or Advanced Automation AI and/or Advanced Automation Examples Claim” Technology Ping An’s “ Smart uses image Tractable is using AI to Zurich Insurance Shift Fast uses AI to find patterns of fraud in deep claims executing - self Axa’s “Fizzy” uses datasets, which can then be applied to incoming claims parametric contracts to uses image recognition and review paperwork (e.g. medical in order to recognition to support triaging processing cost - flag potential instances of fraud compensate customers for reports), speeding up pricing algorithms to and validate repair recognize 16 seconds estimates automatically automotive damage, improving from hours to times delayed flights, eliminating the 17 18 claims efficiency process need for a claims by over 40% Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 99 The New Physics of Financial Services |

101 istribution Insurance | Strategy C: Improve and expand d s trategies C existing distribution channels, of new and AI is augmenting the capabilities reach and scale allowing insurers to expand their W h y d i s t r i b u t i o n m a t t e r s insurance ix channel Simplifying the complexities of m Importance of Opportunities in emerging m arkets By introducing better exploration and search tools Digital and scalable channels are best positioned to into Insurance products and clients are increasingly complex. digital channels, institutions can reduce costs while direct capture the high growth of insurance sales in developing Augmented tools allow agents to manage those needs in a cost - effective way markets customers’ interacting with customers through their preferred channel O b s e r v e d o p p o r t u n i t i e s Increase the efficiency and capabilities of sales agents driven insurance delivery - and experience Use mobile Improve scale efficiencies to enter new markets oint Pain p oint Pain p oint Pain p intensive labour High upfront costs have stymied growth opportunities in - Lead generation and client acquisition is a Insurance purchases are disconnected from the underlying - served markets process with a lot of tedious work traditionally under assets and experiences they protect nsurance markets in countries i For example, expanding into life The best insurance salespeople spend a significant portion of their Currently, insurance policies are largely purchased through a uneconomical due to the need and time generating leads, casting a big net with lots of churn has traditionally been such as India different channel the asset or experience they are protecting, from leads wasting time on cold with a doctor check health an upfront to do requiring customers to go through an unnecessary purchase cycle that is divorced from what is being insured apabilities New c apabilities New c apabilities New c  Non - traditional data can be Advanced visual using party -  Automated decision - Integrations with third  Correlating sentiment and   Predicting lead quality  can with historic usage patterns and points of sale can apps analysing recognition making can provide instant used as a proxy in place of machine learning by , automatically validate official sales data can predict the pricing and underwriting to external data (e.g . sentiment seamlessly integrate ups - check physical doctor reducing the cost of verifying bind documents (e.g insurance purchases into the . medical likelihood of policy quote, analysis on social media) can and issue life insurance application asset purchase at the point of a efficiency real cancellations or renewals personalized policies in forms and doctors’ reports) improve sales sale time AI and/or Advanced Automation AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples Examples Allstate uses image recognition to offer quotes on insurance Cover has launched a joint venture with Chinese insurers to Baidu is using its personal lines agent network to sell use an internal chatbot policies based on a picture taken via the mobile commercial lines data app. It connects to and AI capabilities to develop an AI underwriting powered - products. It has developed 20 engine. This can assets a wide number of insurers for a broad set of insurable that agents can use to provide accurate quotes and valuable cost expand digital originations at low 19 phones advice for complex commercial clients including cars, homes, and pets Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 100 The New Physics of Financial Services |

102 t of | Strategy D : Insure against n ew Insurance ypes risk in new w ays D AI allows agile, enabling them to deploy new institutions to be more products in response to emerging risks n e w r i s k s m a t t e r W h y i nsurance ines speciality of ost Prohibitive c l raditional Evolution of motor risk a compression Margin t cross - property and casualty driving cars will shift - Self Unique, hard markets to to - insure cases have led to an expensive Insurers seeking to maintain existing profit margins will specialty products are protection market that limits access to nsurance commercial i nsurance , leading to falling margins for i need to find new risk areas where insurance for customers insurers saturated and rates are at a premium less O b s e r v e d o p p o r t u n i t i e s Develop modularized policies Use proxy data to insure new risk categories Introduce new pricing and payment models Pain p oint Pain p oint Pain p oint Insurers and Static payment models are not matched to customer needs rely on historical cases when actuaries Insurance policies are predefined and rigid, and cannot be insurance assessing risk probabilities and pricing customized without significant cost Traditional insurance policies restrict users and insurers to single is a lack of historical Insurers have difficulty bundling and unbundling their products as When there data, or when the risk models are upfront payments to cover risk terms and force a renewal period. insurtechs insurance . cybersecurity policies are sold as rigid and fixed entities. This limits the have demonstrated a market desire for not robust enough, risk categories go uninsured (e.g Yet emerging such insurance mile car ) nsurance i specific - product level of customization and modification that can be and more flexible pricing models (e.g . pay - per offered over the - as for laptops) life of the policy apabilities New c apabilities New c apabilities New c   Automated underwriting Dynamic behavioural pricing  damage methods can correlate risk and as proxies for used can be Alternative data sources  Dynamic pricing and allow damage reduces the marginal cost of data to a variety of data feeds (e.g repair bills, sentiment analysis of news) in order to time - series) in real . data (e.g . underwriting models time correlate risk with damage and build actuarial tables , in order to correctly price and underwrite Insurance originating, allowing for different components of a dynamic underwriting at no policies whose prices and coverage vary by usage patterns and policy to be priced, bundled or behaviour sold separately (e.g policies . additional cost (e.g . on an hourly basis) - insurance on a per use basis) AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples Examples AI and/or Advanced Automation allows users to State Farm concept” of “proof has run a Cyence uses AI to quantify the financial impact of cyber policies for insurance miniature purchase competition through Trov - risk by analysing technical and non - technical data sources to understand specific electronic devices, and allows them to turn that coverage Kaggle to develop a program that uses computer vision and photos to identify distracted drivers. This data can be used to micro target at will, offering users the ability to dynamically control cyber - risk of portfolio companies - the on and off 21 premiums lower and offer safer drivers their risk exposure Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 101 The New Physics of Financial Services |

103 complement nsurance i that ervices s on Insurance | Strategy E: Offer add - E make use of unique their internal data and provide AI allows insurers to shelves their complement service offerings product that W h y n e w s e r v i c e s m a t t e r nsurance value changing i techs Large Declining customer engagement propositions Digitization is lowering margins Customer engagement Technology players that integrate insurance into other more difficult as is becoming As more activities in insurance are digitized and margins service offerings are changing the core value of insurance shrink, institutions will need margin to find new high are selling, claims) . traditional touchpoints (e.g - and pushing incumbents to service - in unfamiliar areas and self increasingly automated compete revenue pools if they are to maintain profitability O b s e r v e d o p p o r t u n i t i e s Advise clients on prevention strategies to lower their risk exposures risk their Provide predictive analytics to clients that help them better understand oint Pain p oint Pain p Customers’ limited ability to analyse the effectiveness of risk Customers have limited ability to understand their own risk exposure, which affects their ability - mitigation strategies increases to plan for the future risk exposures and associated insurance premiums prevention measures, such as regular upkeep and behaviour adjustments, can reduce expense - complex task Assessing risk exposures is a Risk for insurers, let alone for their customers. Yet insight into particularly for commercial clients that may ratios for insurers while improving net outcomes for customers, but customers lack clarity about how future risk is a core input into planning and forecasting, changing risk profile need to adapt their business decisions or make major investments based on their proactive measures will impact their risks, costs and activities in the future New c apabilities New c apabilities Personalization at scale allows insurers to time monitoring provides insurers with cosystem analytics using AI allows  E Detailed insight generation  allow insurers to Real  -  tailor advice on how to reduce risk exposure, combine data from their customers, from individualized insights that can be sent to insurers to visualize risks in an intelligible way, allowing them to pass more information customers, enabling them to learn about their suppliers and from the market to deliver in very specific circumstances, that provides generic insights actionable, non targeted advice to customers faster and more behaviours and how associated risk changes to their clients (e.g. the probability of flooding - efficiently in various areas) AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples has Roc Connect uses IoT data from customers’ Zendrive FM Global measures driving behaviour and developed RiskMark, a service is a unique service offering Vitality Manulife that analyses photos and notes to provide homes to enable insurers to offer solutions that produces safety insights that coach drivers to that uses health data from wearables to present for real estate clients on the riskiness of improve behaviour customers with insights into their health risks make homes safer proactively, rather than by insights 22 23 various properties across their portfolios simply being reactive to events and help them track activity Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 102 The New Physics of Financial Services |

104 Conclusion Insurance | Looking forward becomes more personalized, customers will enjoy better prices and coverage. However, this As insurance 1 or excluded by priced out raises challenges for those who are more individualized models time incumbents will start but - AI will push insurance to be more connected, more real and more accurate, 2 large technology firms to data battlefront compared from a position of disadvantage on the policies and Current regulatory frameworks will need to adapt to enable the issuance of dynamic coverage 3 As AI reinvents processes across the insurance value chain, many roles will be displaced, necessitating a 4 plan to transition those displaced into new functions time claims processing will necessitate new approaches to mitigating - Dynamic insurance policies and real 5 the potential for increased fraud Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 103 The New Physics of Financial Services |

105 Insurance | References References BankingTech 1. - the - incredible - growth - of - fintech/54% . Retrieved from “Infographic: The Incredible Growth of Fintech”, as of 5 March 2018. https://www.bankingtech.com/2018/03/infographic - . Retrieved from - services/insurance/ey EY 2014 - global - customer - insurance http://www.ey.com/gl/en/industries/financial survey 2. “2014 Global Consumer Insurance Survey”. - American Association for Justice . Retrieved from https://www.justice.org/sites/default/files/file - “How Insurance Companies Deny, Delay, Confuse and Refuse”. 3. uploads/InsuranceTactics.pdf Reuters . Retrieved from https://www.reuters.com/article/us 4. ping - an - ins - stocks - alibaba - group - “Alibaba, Tencent Chairmen Invest in China's Ping An Insurance”, as 2 December 2014. - - - chairmen - invest - in - chinas - ping - an tencent insurance - idUSKCN0JG0M520141202 tenc/alibaba - “Amazon , Berkshire Hathaway and JPMorgan Chase & Co. to Partner on U.S. Employee Healthcare”, as of 30 January 2018. Business Wire . Retrieved from 5. - Berkshire - Hathaway - JPMorgan - Chase - partner https://www.businesswire.com/news/home/20180130005676/en/Amazon U.S - 6. The Wall Street Journal . Retrieved from https://www.wsj.com/articles/axa - to - buy - xl - group - for - 15 - 3 - billion - 1520236823 “AXA to Buy Insurer XL Group for $15.3 Billion”, as of 5 March 2018. “AIG to Buy Reinsurer Validus for $5.56 Billion”, as of 22 January 2018. - . Retrieved from https://www.reuters.com/article/us - validus - m - a - aig/aig - to - buy 7. reinsurer - validus - for - 5 - 56 - Reuters billion - idUSKBN1FB1FP - “Allianz ompletes 49% Acquisition of UK General Insurer LV”, as of 29 December 2017. Reinsurance News . Retrieved from https://www.reinsurancene.ws/allianz - 8. C 49 - completes acquisition - uk - general - insurer - lv/ 9. “ Insurance Promises Asia Much More than Peace Of Mind”. Financial Times . Retrieved from https://www.ft.com/content/ae46f220 - 1077 - 11e7 - b030 - 768954394623 - 10. “Global Insurance Trends Analysis 2016”. EY. https://www.ey.com/Publication/vwLUAssets/ey - global - insurance trends - analysis - 2016/$File/ey - global - insurance - trends - analysis - 201 6.pdf 11. “Property & Casualty Insurance: 2017 Industry Primer”. Retrieved from http://www.reperiocapital.com/home3/Wordpress/wp - content/uploads/2018/02/MS - PC - Primer - Morgan Stanley. 2017.pdf 12. . Retrieved from https://www.theguardian.com/technology/2017/jan/05/japanese - company - “Japanese Company Replaces Office Workers with Artificial Intelligence”. The Guardian - office workers - artificial - intelligence - ai - fukoku - mutual - life - insurance replaces - “Swiss Re to Work with IBM Watson to Harness the Power of Big Data for 13. Swiss Re . Retrieved from Reinsurance”, as of 22 October 2015. http://www.swissre.com/media/news_releases/Swiss_Re_to_work_with_IBM_Watson_to_harness_the_power_of_Big_Data_for_Reinsurance. l htm it “Insurance You”, as of 10 July 2015. Nasdaq . Retrieved from https://www.nasdaq.com/article/insurance - fraud - and - 14. - Fraud and How It Costs - costs - you - cm494981 how 15. “U.S . Property and Casualty Insurance Industry”. NAIC . Retrieved from http://www.naic.org/documents/topic_insurance_industry_snapshots_2017_industry_analysis_reports.pdf 16. “Zurich Insurance Starts Using Robots to Decide Personal Injury Claims”, as of 19 May 2017. Reuters . Retrieved from https://www.reuters.com/article/zurich - ins - group - claims/zurich - - insurance - using - robots - to - decide - personal starts injury - claims - idUSL2N1IK268 - 17. “Ping An Financial OneConnect Unveils “Smart Insurance Cloud” t o Over 100 Insurance Companies, as of 6 September 2017”. CISION PR Newswire. Retrieved from https://www.prnewswire.com/news - releases/ping - an - financial - oneconnect - unveils - smart - insurance - cloud - to - over - 100 - insurance - compan ies - 300514482.html - 18. AXA . Retrieved from https://www.coindesk.com/axa - using - ethereums “AXA Is Using Ethereum’s Blockchain for a New Flight Insurance Product”, as of 13 September 2017. blockchain - new - flight - insurance - product/ Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 104 The New Physics of Financial Services |

106 Insurance | References References Business Insurance Agents Speed Up Quoting with Context Sensitive Help System”. Early . Retrieved from http://www.earley.com/knowledge/case - 19. “Allstate - intelligent - agent - reduces - call - center - traffic - and - provides studies/allstate%E2%80%99s help - 20. “Chinese Search Giant Baidu Forms New Online Car Insurance Business”, as of 8 June 2016. South China Morning Post . Retrieved from https://www.scmp.com/tech/china - tech/article/1968770/chinese - search - giant - baidu - forms - new - online - car - insurance - business “State 21. Farm Distracted Driver Detection”. Kaggle . Retrieved from https://www.kaggle.com/c/state - farm - distracted - driver - detection - 22. Global Uses Data Science to Reduce Client Risk”, as of 9 December 2014. Fortune . Retrieved from http://fortune.com/2014/12/09/fm - global - data “How Commercial Insurer FM science/ 23. “The Manulife Vitality Program”. Manulife . Retrieved from https://repsourcepublic.manulife.com/wps/portal/Repsource/SalesResources/Insurance/ManulifeVitality/!ut/p/z0/04_Sj9CPykssy0xP LMn Mz0vMAfIjo8zivQx9TTwc3Q18DPwsLA0cXQID PSzC3A0svA30C7IdFQGTGV1D/?WCM_GLOBAL_CONTEXT Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 105 The New Physics of Financial Services |

107 Payments Sector exploration

108 overview Payments | Sector Payments service providers face sustained margin pressures that threaten their traditional business models O v e r v i e w I s s u e s f a c i n g t h e s e c t o r Sector description and chapter s cope Shrinking margins due to 10% decrease in debit card interchange fees − in the US 2 from increased competition and 2013‒2017 merchant pressure institutions that facilitate the exchange of money between all The payments industry includes (e.g. parties, across all channels cheque versus card). category of APIs − The second - fastest - is for growing Growing complexity and This chapter focuses on the end - to - end payments value chain, from payee to payer; this chapter services; APIs increase difficulty in ensuring payments and financial also includes dialogue on the underlying payments infrastructure supporting the flow of money on a 3 interoperability fragmentation and interoperability complexity national and global level. − 271 mobile money solutions existed across 93 countries etworks n payments International , interoperable as of 2015, with very few being within even 4 funds ) country the same International t ransfer (e.g anking b . correspondent in most important factor second “Faster service” is the − Increasing customer 5 expectations for instant institutions retaining customers, according to payments n etworks Domestic payments payments Instrument Payment Merchant Merchant Issuer − believe they will be providers payment Only 51% of UK Payee Payer issuer processor network processor acquirer 6 time payments globally by 2020 able to send real - : Cash, cheque, credit card, debit card, prepaid, ACH, money transfer, other Method in Disintermediation as a expected from P2P payments − 200% growth 7 globally 2016‒2018 result of P2P payments trends Sector . PSD2 efforts Payments modernization − , UK Open (e.g Banking Standard) are opening access to payments → value chain and enabling unified mobile Open data flows are compressing the payments infrastructure to third parties payments experiences has led to a focus on premium cards in Increasing competition on rewards and benefits → 1 in fighting 53% of institutions report increased costs − Growing burden of developed markets 8 financial crime and fraud economic crime over 2014‒2016 via mobile channels has dramatically expanded the presence of digital Increasing accessibility → 246% increase in global credit card fraud loss from − payments in emerging markets 5 2012‒2017 based payment systems raises uncertainty about the future of → Innovation in blockchain - − US institutions were fined $12 billion for failing to meet networks payment centralized 9 financial crime requirements from 2009‒2015 Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 107 The New Physics of Financial Services |

109 Payments | Strategies summary AI presents new tools to fight fraud, respond to the shifting form of payments and draw valuable insights from data ayments : p changes in The rise of AI will initiate and accelerate the following - Payments customer experiences time payments Data becomes the most valuable The risks of real shift away from An accelerating cash led - aspect of payments continue to disappear in favour of mobile payment as as new AI are overcome businesses AI removes the need for manual effort detection interfaces, in all markets, as digital pattern - as AI unlocks opportunities to create methods make new insights significant inroads against financial with are bundled payments automated . from customers (e.g - value fraud detection and handling) time fraud) - real (e.g. crime services added include: - Key AI payments enabled strategies in Strategy B Strategy A Strategy C ctivity i alicious m a a fraud Reduce n nd f p oment m disappearing Respond to the ayments Unlock the p ower o ystem s payments the of nsights payment i B A C E.g. → se . → E.g . u E.g → image recognition to Use machine learning to payment providers using payment - data to provide machine learning significantly reduce false positives in - sale authenticate and transact point - of “as a payments merchant analytics fraud detection based service” highlight key examples detail The following slides will explore each strategy, the components and Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 108 The New Physics of Financial Services |

110 nd s payments Payments | Strategy A: Reduce fraud a ystem m alicious a ctivity i n the A AI is increasing confidence when making payments by enabling customers’ time interventions and responses to criminal activities - real m a t t e r s W h y r e d u c e d m a l i c i o u s a c t i v i t y Emerging Speed of payments cyberthreats Growing, high - cost processes As regulators increase security and privacy requirements Institutions’ lack of confidence in their ability to combat the Institutions must invest in new methods for protecting the increased velocity and volume of fraud has slowed down (e.g payments system to keep pace with the sophistication of GDPR), the cost of processing payments for . will grow institutions time payments - the implementation of real cybercriminals O b s e r v e d o p p o r t u n i t i e s time surveillance capabilities Deploy real to eliminate false positives precision Increase detection Automate compliance and reporting - oint Pain p oint Pain p oint Pain p Back Costs of resolving and reporting on criminal activity Institutions lack resources to identify the breadth of threats - office resources are plagued by false positive results resources analysing and auditing false are spent methods predominantly place the onus on Traditional “flagging” System limitations and data inaccuracies that require manual entry on ignificant S - based systems. These systems and reporting responding slow down the process of understanding, positive results generated by rules manual review, limiting the ability to screen a high volume and increasing staff levels to - on malicious events. This requires ever the root cause, provide limited risk - and ordering insight into increasing velocity of transactions. As a result, institutions have placed dollar thresholds on the screening of transactions and have - response required vis à meet regulatory and customer expectations vis flagged transactions - 10 risk slowed the speed of settlement to combat apabilities New c apabilities New c apabilities New c can reporting Complex relationship Automated  Looking for patterns in  using gathering  -  End - to - end resolution engines can use machine learning to data Dynamic  generate detect patterns in transactions and proactively intervene with source, sort, and AI to validate and transmit using machine mapping (structured datasets new can recognize learning more increased confidence to stop abuse before a transaction is store reporting and unstructured) can transaction details enables the provide precise detail and requirements to maintain complex patterns, as well as - automation of downstream back settled. This can allow for threshold limits to be expanded while and understand results from past logs abuse audit trails, risk office processes and streamlines context on the potential for limiting losses due to workflows abuse investigations reports AI and/or Advanced Automation AI and/or Advanced Automation Examples Examples AI and/or Advanced Automation Examples generates fraud insights by integrating data from across uses AI automatically catalogues all business data and indexes it Alation FICO’s Falcon Platform - driven predictive analytics to Feedzai services to precise hyper for easy reference, using machine learning to learn the lexicon of risk learns from institutions. Falcon - detection - fraud provide profiles. It also uses machine channels to generate real time case dispositions in smart detects 61% real time. learning to process transactions in business. Alation also proactively makes the owned data - and integrates merchant Feedzai 11 - reporting processes rates into continuously updated “behavioural profiles” for every individual to support more fraud earlier and recommendations without increasing false positive compliance Key Findings Capt. - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Cross mkts . Mkt. infra . Wild - Card Scenarios Sector Explorations 109 The New Physics of Financial Services |

111 o p Unlock the p ower Payments | Strategy B: f i ayments nsights B AI enables payments providers to generate new revenue streams by using provide unique insights to datasets their W h y p a y m e n t s d a t a m a t t e r s value data Source of unique customer insights Declining margins demand diversification - Access to excludable, high in - time and flow - based, - Participants across the payments value chain Payments information is point Margins in payments may continue to decline if regulators continue to impose interchange fee caps and customers (e.g allowing data holders to develop . issuers, retailers) are seeking access to payments precise analytics that data to better understand customer needs expect fee free solutions provide unique insights capable of being put into action O b s e r v e d o p p o r t u n i t i e s Offer Act as the ultimate personal shopper for customers to merchants prediction “as a service” Create an advisory capability for macroeconomic trends Pain p oint oint oint Pain p Pain p Time Many institutions have insufficient data - analytics capabilities Traditional economic data is often generic and infrequent - consuming data collection and analysis their sized business in particular struggle to keep - is Small and medium Macroeconomic data (e.g It is difficult for customers to make optimal decisions based on spending) demographics, consumer . - value usually produced infrequently (e.g routine and high ), is not quarterly or annually pace with rising data flows, affecting their ability to draw insights, . personal financial constraints, for both . granular, nor provided in machine - transactions, due to the time and effort required to source readable targeted take action or make meaningful decisions (e.g formats. This limits the - ability of third parties to derive deep, product information promotions) comparative specific insights apabilities apabilities apabilities New c New c New c Bespoke Optimization engines can that  s analytics  olutions can combine payments Build dashboards   Analysis of can Advanced analytics  automatically generate core party datasets third (e.g. capabilities to learning - - machine unique datasets with providers’ generate highly specific and use deep learning and macroeconomic indicators and provide targeted insights data, available and other clients. This personal for merchants accurate recommendations by visualization to allow users - time statistics to in turn can provide contextual and specific recommendations to drill into data in a simple market can using machine learning to information) provide real and personalized requests - to - use format various client satisfy and easy products next best action) for services, . (e.g insights into the and predict shoppers’ analyse provide shoppers promotional campaigns. needs of behaviour AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples PXP Solutions’ AnyPay provides Dynamic Yield Mastercard’s SpendingPulse uses near - uses AI to analyse the Upserve supplies merchants real - time purchase data Spotify - personalized recommendations provides smart analytics that to build customer reports on macroeconomic trends on up to a listening habits of its with data marketing , analytics million users to generate to shoppers that maximize allow smarter resourcing and and loyalty tools built on credit industries and geographical weekly basis for a wide variety of 100 13 12 areas, from retailers to national governments playlists of suggested marketing decisions card transactions songs revenue for leading retailers Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 110 The New Physics of Financial Services |

112 of Respond to the disappearing m oment Payments | Strategy C: payment C AI provides valuable tools for payment providers to preserve relevance as disappears the moment of payment W h y s e a m l e s s p a y m e n t s m a t t e r Payments are becoming automatic background events influence adoption Choice of payment will be made by algorithms Ability to - Payment providers are - pressed to differentiate payment process As platforms and invisible point hard of - sale payments solutions the skip over Modern commerce solutions meaning fewer take hold, the decision about themselves in ways that can improve the “stickiness” of (e.g . order), - which payment method to use to - click - Amazon one interactions between customers and their cards will increasingly be made by optimization algorithms customers as the point of payment disappears O b s e r v e d o p p o r t u n i t i e s Compete to become a provider of invisible payments infrastructure Drive usage of payment products by offering bespoke incentives and rewards Pain p oint Pain point among Payment providers risk losing control of the merchant environment as “invisible commerce” card providers - The app ecosystem is winner takes creating a - all environment becomes more important decision systems, customers are making fewer choices - Whether through digital wallets or support Large technology players are introducing radically new point - sale solutions for merchants, moving about which card or payment method to use at a point of transaction. Instead, they are tending to use of - onboarded the transaction below the level of perception. If widely adopted, invisible payments could eliminate the one of a few cards that have been onto these payment platforms (e.g Uber and Apple . Pay). decisions interaction between payment providers and customers, perhaps even replacing traditional card This is limiting the ability of card issuers to influence customers’ top - of - wallet networks with new payment rails apabilities apabilities New c New c Automated checkout can be delivered “as a can be Real  Seamless authentication can be introduced - time benefits optimization can   Niche and custom reward offers  by generated using machine learning, optimizing by payments providers using image service” intelligently maximize benefits for customers providers using image payment carts travel or based events such as - through time recognition and/or biometrics to verify the reward rates for different purchasing recognition to compile shopping and customer’s promotions unique determine when to collect payment, removing identities without disrupting the customer categories based on each purchasing patterns experience the “point of payment” as a separate event that requires conscious effort from customers AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples uses facial recognition as Amazon Go is a cashier - less store that tracks Alipay’s Smile to offer customers real time and American Express in Mastercard accesses data from Visa, Mogl to Pay what visitors have put in their cart and a method of authentication and consent, offering how busy the restaurant nearby restaurants (based on personalized, time at based cashback rewards - preferred menu automatically charges promotions etc.), optimizing benefits for both customers and restaurants them using their retail customers a frictionless checkout is, the store method when they exit payment experience Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 111 The New Physics of Financial Services |

113 Conclusion Payments | Looking forward Data is quickly emerging as a critical source of value in the payments sector, raising challenging questions 1 around ownership and privacy frameworks will be The complexity of today’s fragmented payment networks means automation increasingly 2 to facilitate necessary interoperability Collective efforts to combat payments fraud and other abuses through the pooling of data would serve the 3 system safety and security of the broader financial interests of individual institutions and improve the AI - management tools will make real - enabled AML and fraud - time domestic payments safer, but regulations 4 border data flows may constrain faster international transfers - on cross wide outages and stability and the potential for broader contagion, the In order to mitigate the risk of system - 5 rails reliability of new payments infrastructure must be validated before it fully displaces legacy Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 112 The New Physics of Financial Services |

114 Payments | References References II (Debit Card Interchange Fees and 1. . Retrieved from https :// www.federalreserve.gov/paymentsystems/regii - average - interchange - fee.htm Routing)”. “Regulation Federal Reserve :// , Financial, Analytics API Categories See Big Programmable Web . Retrieved from https Growth as of May 21, 2018”. www.programmableweb.com/news/financial - ecommerce - “ 2. Data enterprise - categories - see - big - growth/research/2017/04/10 - api “Solving access”. World Bank . Retrieved from https :// blogs.worldbank.org/psd/solving - payments interoperability for universal financial - interoperability - universal - financial - access 3. payments “Redrawing the lines: FinTech’s growing influence on Financial Services” . PwC . Retrieved from https://www.pwc.com/jg/en/publications/pwc - global - fintech - report - 17.3.17 4. final.pdf - 5. ACI Worldwide . Retrieved from https ://www.aciworldwide.com/ - / media/files/collateral/trends/the - road - to - faster - payments - on - demand - webinar - “The Road to Faster Payments”. infographic.pdf 6. . “What the Introduction of a Faster Payments Scheme in the US Means for Payment Service Providers”, as of 22 June 2018. Paysafe . Retrieved from Linden, Todd - - - introduction - of - a - faster - https://stories.paysafe.com/news/what - scheme - in the the - us - means - for - payment - service - providers / payments 7. “Global P2P Mobile Payments Value Set to G row by 40% in 2017”, as of 10 May 2017. Payments Cards & Mobile. Retrieved from https ://www.paymentscardsandmobile.com/global - p2p - grow mobile - payments - value - set - - 40 - 2017 / 8. “2017 Retailers”. ACI Worldwide . Retrieved from https ://www.aciworldwide.com/ - / media/files/collateral/trends/2017 - global - payments - Global Payments Insight Survey: Merchants and - - merchants - and - retailers.pdf insight survey Stabe , Martin, and Aaron Stanley. “Bank Fines: Get the Data as of July 22nd, 2018”. Financial Times . Retrieved from http:// ig - legacy.ft.com/content/e7fe9f25 - 542b - 369f - 83b2 9. - 5e67c8fa3dbf McLaughlin ’ as of April 21st, 2017”. PaymentsSource . 21 Apr. 2017 . Retrieved from 10. , David. “Artificial Intelligence Can Cut Money Laundering’s ‘acceptable Losses :// www.paymentssource.com/opinion/artificial - intelligence https can - cut - money - launderings - acceptable - losses - 11. “Modern Payment Fraud Prevention at Big Data Scale”. Feedzai . Retrieved from https :// feedzai.com/wp - content/uploads/2015/10/Feedzai - Whitepaper - Modern - Payment - Fraud - Prevention - - Big - Data - Scale.pdf at “Spotify expects nearly 100 million paid users by year’s end... and more losses as of March 28th, 2018”. The Industry Observer . Retrieved from 12. https ://www.theindustryobserver.com.au/spotify - expects - nearly - 100 - million - paid - users - by - years - endand - more - losses / - 13. MEDICI. Retrieved from https ://gomedici.com/10 - payments “10 Payments Companies Competing in Analytics as of January 14, 2015”. companies - competing - analytics / Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt . Capt. mkts . Mkt. infra . Wild - 113 The New Physics of Financial Services |

115 Investment management Sector exploration

116 management | Sector overview Investment Investment managers are adapting their customer experience and product offerings in response to new competition O v e r v i e w I s s u e s f a c i n g t h e s e c t o r 3 description s cope chapter and Sector 62% xpectations of − Customer e of consumers find cross - channel switching important igital channels d are to a person − speak say they online services only when 48% increasing provided by a broad Investment is a diverse sector made up of a variety of services management 3 enough aren’t good investment management firms will offer a combination of Typically, group of financial institutions. ) . direct investing), brokerage (i.e. asset management facilitating transactions (i.e. and advisory (i.e financial planning) services. 51 − p alent t Ageing adviser ool States, in the United advisers is the average age of financial scope Chapter and this is set increase in the coming years as the industry to 4 Flow of funds struggles to recruit younger advisers managers Asset Capital Investors Asset owners markets of millennials are open to trying financial Two - Risk of new e ntrants − thirds Wealth E.g E.g Individuals, . . . ETFs, hedge E.g managers , investment f - robo corporations / , unds services from trusted brands (e.g o and wning customer Nike, . Google 5 e commercial advisers ntities pension f , unds institutional , experiences Apple) a unds f sset m anagers fee compression is expected over the next 8% revenue and three Advisory − years, fee 6 50% fall in forecasting up to a with analysts fees c ompression trends Sector fee active to low investments. → Shift from high - s - fee passive A consumers and institutions 7 − $50 trillion in global wealth is u return global Large nmanaged earning no fee - conscious become increasingly , assets under management (AuM) are flowing to low - fee eposits d passive investments − $17 trillion of potential new AuM in China by 2030 (14% 8 CAGR) → of scrutiny wealth Regulatory Financial Conduct Authority’s review of wealth . managers (e.g 1 managers’ suitability practices in the ) UK traditional particularly investment strategies , → Saturation of strategies, discretionary investment − 63% increase in the share of systematic trading strategies Growth in demand for has reduced profitability due to overuse 9 nvestments alternative as a percentage of all hedge funds from 2011‒2016 i Intergenerational wealth transfer → in developed markets will move trillions of dollars in assets 10 AuM in quant funds $940 billion − 2 America – ) across customers ($ 30 trillion over the next years in North 40 30 Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 115 The New Physics of Financial Services |

117 Investment management | Strategies summary to adapt their business models by altering or enabling investment AI is anagers m replacing core differentiating capabilities changes in investment m anagement : The rise of AI will initiate and accelerate the following Wealth management in emerging Investments will be increasingly Passive products will develop Alpha - seeking arbitrageurs will personalized markets will rapidly develop as investment firms active characteristics as models as as be pushed to new horizons existing mimic complex strategies or develop bridges ingest new data about their digital distribution systematic investors make greater customers gaps their own use of advanced data science and alternative data sources - anagement include: enabled strategies in investment m Key AI Strategy D Strategy E Strategy B Strategy A Strategy C expand Enhance and Become a Offer m ore c ustomized hyper Pioneer e merging - arkets efficient, and ealth w Use data to generate a lpha m ealth r eturns w income - low and ortfolios p differentiate i a dvisory anager m nvestment i fee - low nvestment B A E D C new data sources to se clients with a . provide E.g → deep learning and u . E.g → se machine learning to → E.g . u digitized account setup se u . E.g → u . E.g se → cutting macroeconomic analysis and management to expand techniques make other better inform and articulate - edge branded chatbot that to innovate in the creation of - worth access for lower - net than cheaper faster and investor profiles and seamlessly integrates with traditional methods clients preferences investment strategies existing advisory relationships key examples highlight components and detail the The following slides will explore each strategy, Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 116 The New Physics of Financial Services |

118 | Strategy A: dvisory a ealth w Investment management expand Enhance and A advisers AI is allowing wealth to provide personal and targeted investment advice to effective manner - customers in a cost market - mass W h y i m p r o v e d a d v i s o r y m a t t e r s become self Pressure to service Cost of service Demand for d ifferentiated e xperience - AI and automation solutions have the potential to advice can serve as a management - wealth Improved service tools become more sophisticated and easy As self - are value their are seeking to change customers through reduced decrease the cost of serving mass advisers to use, differentiator for market customer segments that - advisers by served human capital costs and faster timelines - traditionally under proposition for customers O b s e r v e d o p p o r t u n i t i e s insights highly personalized with advisers Equip economic insights detailed Share more investments users to effectively manage their Enable Pain p oint oint Pain p oint Pain p - Advisers struggle to differentiate advice and retain customers Impulsive and uninformed decisions result in poor outcomes counter financial advice is limited in scope the - Over advisers advisers are unable to provide Mass - market Branch . are ubiquitous but limited in their ability to serve With limited access to data, wealth investors sometimes fall into “beginner traps” (e.g customers with tailored financial advice due to the size of their detailed and specific buying high and selling low, impulse investing, tax inefficiencies), financial advice, resulting in generic advice age, advisers. This, in turn, leads to and a lack of differentiation among books. This gap is often extended by other disconnects (e.g because the oversight afforded to higher . - net - worth clients such – advisers as private bankers is not available low customer stickiness clients and their – lifestyle) between apabilities New c New c apabilities apabilities New c  Advanced analytics dashboards nalysis detailed insights can use machine learning to look provide   Econometric indicators can combine economic datasets and Cross - product a clients’ needs and enable easy calculations to optimize about market events to provide customers with relevant insights and automatically and financial products customer’s across a areas of optimize data on macroeconomic trends; customers will tend to stay with expand the branch suboptimal savings . products, services and advice. This can improvement (e.g that houses their historic advis e r role by enabling junior and non - dedicated financial data and consequently allocations) the platform advice to – and potentially niche – to provide personal advisers delivers stronger insights modelling) clients ( e.g. portfolio AI and/or Advanced Automation Examples Examples AI and/or Advanced Automation AI and/or Advanced Automation Examples Right Capital - offers software to has introduced an Albert combines customers’ financial data across insurance, ForwardLane Addepar provides easy - to use financial - planning software wealth employees in managers that integrates deposits, lending and wealth management to provide personalized API through which investors can tools services . tax projections) to aid financial (e.g customers’ financial information performing ad hoc analysis that helps clients make decisions and query the firm about how financial advice that considers the individual’s holistic financial strategies finances to plan their position and wider market - generate detailed and specific related events affect their portfolios can in real time reports Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 117 The New Physics of Financial Services |

119 | Strategy B: hyper anager m nvestment Investment management - Become a i - efficient, low fee B growing management portion of investment AI is taking on a responsibilities cost service at a lower quality - high , delivering f e e m a t t e r s W h y b e i n g l o w - eers uccess q uality or f performance benchmarks Changing s p gainst a ompetition c price Maintain Win It is critical for hedge funds and other high institutions Combining high efficiency with low fees allows end investment - Passive strategies have changed the basis of competition from to among investment m anagers - adjusted return performance “generating alpha” to reduce expense ratios and remain competitive in an products to maintain risk anaemic environment of low interest rates and fees” growth they engage in cost while - cutting low “having O b s e r v e d o p p o r t u n i t i e s office compliance costs Control ballooning back - datasets Mimic advanced strategies while controlling costs Establish passive products that track new Pain p oint Pain p oint Pain p oint Reviews and disclosures are performed manually Funds with specialized strategies tend to have high fees tracking established Passive products are limited to indices Firms that attempt to differentiate based on financial performance Current passive products are built same set of assets and on the Investment institutions have often implemented tactical stop - gap costs operating fund’s easy to analyse and readily available. Many new data must invest heavily in talent, increasing the data, which are solutions using manual processes and paper documents to meet requiring regulatory requirements. This results in sources are unstructured, requiring cleansing and normalizing, which charge higher performance fees it to high operating costs and and unappealing - intensive work with - for more labour each regulatory and thus intensive is labour the need change New c apabilities New c apabilities New c apabilities unstructured data Image Automated legal  - Data  voice, text, images) at scale . (e.g  Parsing  learning - machine through Automation of data analytics  gathering increases recognition can the and using machine disclaimers learning can open up new data sources that can automation can technologies and cloud processing significantly be employed to analytics throughput of talent, allowing a small number of be indexed to market trends. Trading strategies based on these boilerplate text can be used to digitize be generated compliance indices offer differentiated return profiles at lower costs professionals to replicate what previously took an army of find information to generate reports documents and analysts using machine extract key figures learning AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples Fundapps Fortia Financial Solutions tracks position limits, Wealthfront Vortexa, iSentium, fund to mimic parity - risk a has launched RavenPack and many others are packaging Bridgewater’s automates compliance . alternative data (e.g disclosure requirements Twitter feeds or satellite data) as trading “all - weather fund” without the $100 million account and extend differentiated return minimum and lower fees, allowing it to for hedge funds, suggesting a strong demand to use this investment restrictions to create signals processes for fund managers 11 traditional - warnings and regulatory filings its non profiles to base data for new investment strategies client Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 118 The New Physics of Financial Services |

120 more c Investment management | Strategy C: Offer p ustomized i nvestment ortfolios C - AI enables more tailored driven personalized portfolio management outcomes investment customer experiences and better W h y c u s t o m i z a t i o n m a t t e r s Potential to reduce attrition via product Differentiated utcomes for clients Extend a unique c ustomer e xperience financial a vailability o ETFs become more beta - Customized advice and portfolio management ensures smart As new products such as Customization allows institutions to position themselves understanding and more caring popular, stagnant investment managers risk losing that personal finances are aligned to the objectives of as the smarter, more money managers the client customers and AuM O b s e r v e d o p p o r t u n i t i e s understand investor preferences Holistically in real time Develop unique strategies to customize portfolios Offer outcome - based portfolio modelling oint Pain p oint Pain p oint Pain p Current methods for understanding product Portfolio optimization is investment - - customer Existing investment strategies are simplistic and merely instead of focused customers’ between goals and outcomes adjust asset allocation across a limited product shelf focused , leading to needs are inaccurate and onerous for a mismatch customers Often, wealth managers separate customers into preset risk Customers are presented with simplistic - Building customized hedging strategies or risk exposures is a high risk - profile surveys that do - standardized portfolios; this model is easier for skill investment professionals to tolerances, cost activity that requires high that fit buckets not develop nuanced, quantitative profiles of their make value judgements. This means that institutions are not able misalignment with customers’ risks but investment manager does wealth that their leading to individuals feeling managers goals not truly understand their needs financial to offer these services to some or all of their clients apabilities New c New c apabilities New c apabilities  can Modular offerings  customer views Integrating Analysing new data  new Analysing  Building  new Correlating  Achieving continuous  data provide goal by personalize financial can driven solutions sources monitoring in real time with market risk exposures , including - personalized data can develop proxy feeds that - performance using advice in the context of a allocation driven rather than product party - building asset - transaction and third using machine models bundles monitor broad set of potential client life clients’ finances and learning, machine learning data, portfolios for , despite common risk areas allows development immediately when detect goals (e.g. holidays, starting a allows for the generation of their complexity, to of customized personalized rebalancing is necessary (e.g business) expand products real estate) . risk factors AI and/or Advanced Automation AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples Examples strategies CleverNudge combines bot quantitatively tracks Clarity AI - based and human Fountain Money based advice to Fidelity and BlackRock’s Decipher Finance analyses uses machine - cost investors - products give low learning to generate contextual clients that focus on the responsibility of firms, the social build bespoke investment plans for data to reveal investment needs access to strategies such as to and provide which can be used by fund tailored advice triggers and customer specific life goals of the investor segments multifactor weighting and have wealth clients managers to optimize socially 12 responsible portfolios over $100 billion AuM Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 119 The New Physics of Financial Services |

121 arkets ealth management | Strategy D: Pioneer emerging m Investment and low - income w D income markets in a cost - effective AI enables institutions to - serve low manner W h y n e w m a r k e t s m a t t e r developed high Opportunities in arkets m Growth in growth m - arkets Lower switching c osts Digital tools are making it easier for investors to change the Increasingly, small deposit amounts can be Divergent global growth rates promise a boom in profitably managed, allowing for continued growth in wealth of emerging markets, where existing wealth - their investment and advisory providers, creating the management solutions for firms to opportunity mature markets access new customers do not meet customer demand O b s e r v e d o p p o r t u n i t i e s Seamless account setup and portfolio construction Explore digital platforms for distribution existing Digitize customer servicing oint Pain p oint Pain p oint Pain p today accounts is a complex process the cost of Opening investment The cost of opening and servicing small - Dependency on human advisers increases cap accounts operating channels outweighs the potential revenues and regulatory Legal opening an requirements mean that Inflexible and unresponsive user experiences limit digital channels It is often uneconomical for institutions to target low worth - involves completing and signing - investment account often net This forces customers to dig for documents. numerous forms and segments even though their accounts in emerging middle - to a predefined set of actions and recommendations, requiring class investors to access high status of other wealth is significant and rapidly growing person channels for - information that is not readily available (e.g phone or in cost - . person - in investments), as well as make time to be present personalized answers or recommendations apabilities New c apabilities New c apabilities New c allows  Voice search allows clients end automation Data normalization adapt  Building integration points Intelligent dashboards  Digital identity solutions   End - to -  ask questions using natural institutions to access data through - that allows account setup and to every interaction to can be bolstered by AI using allows straight servicing from disparate sources, language with their be connected to image recognition to reduce advisers have to processing of accounts when they do not uncertainty of identity digital platforms that hold customers to make critical opened on digital channels, at increasing the scalability of have certainty over where to verification low marginal cost deposits (e.g information accessible find a particular answer digital channels . payment apps) AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples is built into the Alipay platform and allows Alipay Onfido uses image recognition to verify ID documents, as well as Companion is a pilot project, developed with UBS Yu’e Bao FaceMe , which - – to facilitate cross management clients to pose any economic question allows wealth recognition facial border ID verification of AuM customers to invest their leftover balances. This has led to a virtual avatar of the firm’s chief investment office and receive four $235 billion in to years, making it the increasing the global scalability of onboarding world’s largest money 13 14 conomist market fund answers from UBS’s e chief Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 120 The New Physics of Financial Services |

122 | Strategy E: to r differentiate Investment management eturns Use data and generate a lpha E be used to generate products with new return profiles that are AI can established strategies with uncorrelated W h y d i f f e r e n t i a t e d r e t u r n p r o f i l e s m a t t e r clear d ifferentiator Fund performance Arbitrage is increasingly f leeting Traditional investing strategies are saturated is a As data is commoditized and algorithms operate at high Overall fund (e.g. ability to earn alpha) It is increasingly difficult to generate alpha and identify perceived skill as a skill will investors and enhancing this speed, the window to profit from new strategies shrinks, is a core factor for investment opportunities with high potential through typical strategies result in capital inflows forcing arbitrageurs to be more innovative O b s e r v e d o p p o r t u n i t i e s Analyse vast quantities of data at scale Find new relationships using advanced analysis methods Continuously source new and exclusive datasets Pain p oint oint Pain p oint Pain p Defending proprietary datasets is increasingly difficult Most funds correlated with each other The capabilities of data scientists are limited by imagination, ’ returns are closely creativity and the volume of data that can be parsed the basis It is difficult for institutions to differentiate Across many financial markets, it is increasingly difficult to themselves on of data as it becomes commoditized by service - generate alpha through traditional investment strategies. Markets Many market correlations are non linear or convoluted, making providers. are becoming increasingly efficient as more investors make use of them difficult to analyse However, the exponential growth of data suggests that there is an and turn into useful signals. Traditional a consuming time broad suite of sophisticated investment approaches and analysis techniques are also opportunity in continuously sourcing new and excludable data costly - New c apabilities New c apabilities New c apabilities . deep earning ) to identify l  Use cutting - edge  Automate the collection and algorithms (e.g modern data Employ  purpose - Develop general  Parse unstructured data  to support advanced learning previously unexplored patterns and correlations that support structuring of data - to using storage architecture that analysis technologies investment can derive insights from a algorithms to continuously decision - making to generate alpha for investors - efficient investment decision make large datasets making (e.g . by automatically and economically build new searchable wide variety of types of data accessible (i.e . datasets that can support summarizing key insights) social, quantitative) and . (e.g and sortable) investment analysis audio, text) . formats (e.g AI and/or Advanced Automation Examples Examples AI and/or Advanced Automation AI and/or Advanced Automation Examples WilmotML offers asset managers exclusive datasets and an Eagle Alpha uses machine learning to offer advisory and investment are hedge funds Two Sigma and Renaissance Technologies differentiate their return profiles using quantitative methods that analytics platform to assist in the analysis of alternative data tools, using emerging secular trends (e.g. connected cities, 15 and have . across 24 categories (e.g mobile app usage, satellite and weather, AuM each designer genes) to develop a deep understanding of the macro attracted $50 billion reviews and ratings) environment and inform investment decisions Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 121 The New Physics of Financial Services |

123 Conclusion Investment management | Looking forward cash to increase individual wealth globally, but “dead” There is interest in capturing unmanaged and 1 supervisory bodies must ensure is coupled with customer education and informed consent this The consolidation of institutions and algorithms will deliver lower prices for customers due to automation 2 it and price competition, but will also create new systemic risks in the event of failure Wealth advisory will increasingly become the centre of customers’ financial lives as it gains access to 3 products of financial increasing volumes of data, allowing it to expand reach and take control Participants must consider the risks associated with poor algorithmic decisions and the potential exposure 4 that comes from an increasingly centralized infrastructure for investment management (e.g. shared algorithms) advisory algorithms can The shape of conduct risk will change given that individual mistakes by wealth - 5 impact large sets of clients, while mistakes by individual advisers are more frequent but also more contained Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 122 The New Physics of Financial Services |

124 Investment | References management References “TR15/12 Management Firms and Private Banks: Suitability of Investment Portfolios”, as of 13 October 2016. Financial Conduct Authority . Retrieved from 1. : Wealth :// www.fca.org.uk/publications/thematic - reviews/tr15 - 12 - wealth - management - firms - and - private - banks - suitability https capitalizing 2. Intergenerational Shift in Wealth”. Accenture . Retrieved from https :// www.accenture.com/us - en/insight - Capitalizing on the - intergenerational - “The ‘Greater’ Wealth Transfer: shift - wealth - capital - markets - summary 3. “Customer Experience: Innovate Like a FinTech”. EY . Retrieved from http ://www.ey.com/Publication/vwLUAssets/ey - gcbs - customer - experience/$ FILE/ey - gcbs - customer - experience.pdf all 4. of Advisors Nearing Retirement, Says Cerulli ”, as of 17 January 2014. Financial Advisor . Retrieved from https ://www.fa - mag.com/news/43 -- of - % - advisors - are - approaching - “43 retirement -- says - cerulli - 16661.html 5. “The Future of Customer Experience in Financial Services”. Akamai . Retrieved from https :// www.akamai.com/us/en/solutions/for - cios/the - future - of - customer - experience - in - financial - services.jsp Management. “Tackling the Cost Challenge – Latest Trends & Opportunities”. Morgan Stanley Global Asset 6. 7. “$50 Trillion of Cash on the Sidelines Could Be Good News for Stocks and Gold”, as of 5 November 2016. Business Insider . Retrieved from http://www.businessinsider.com/50 - trillion - - of - cash - on - the - sidelines - good - news - for 11 stocks - and - gold - 2016 - 8. “ Assets Under Management to Hit $17tn by 2030 ”. Financial Times . Retrieved from https :// www.ft.com/content/34be24c4 - c3ae - 11e7 - b2bb - 322b2cb39656 China’s “The / Funds”. Novus . Retrieved from https ://www.novus.com/blog/rise - quant - hedge - funds 9. Rise of Quantitative Hedge 8713 “Quant $1tn Management Mark”. Financial Times . Retrieved from 10. :// www.ft.com/content/ff7528bc - ec16 - 11e7 - Hedge Funds Set to Surpass - 513b1d7ca85a https 11. Just Got Better with Wealthfront”, as of 22 February 2018. Wealthfront . Retrieved “Investing https ://blog.wealthfront.com/risk - parity / from 12. “Smart Beta Funds Pass $ 1tn in Assets”. Financial Times . Retrieved from https :// www.ft.com/content/bb0d1830 - e56b - 11e7 - 8b99 - 0191e45377ec 13. “Ant Financials Yu’e Bao Caps Daily Investment at $3,000”, as of 7 December 2017. . Retrieved from https :// www.reuters.com/article/us - ant - financial - fund - regulation/ant - Reuters - financials - bao - caps - daily yue investment - at - 3000 - idUSKBN1E11FQ - 14. “AI and avatars set to alter wealth management”, as of 4 July 2018. Financial Standard . Retrieved from https://www.financialstandard.com.au/news/ai - and - avatars - set - to - alter - wealth - management 123160588?q=UBS - 15. “ Two Sigma Rapidly Rises to Top of Quant Hedge Fund World”. Financial Times . Retrieved 4a9c83ffa852 https :// www.ft.com/content/dcf8077c - b823 - 11e7 - 9bfb - from Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 123 The New Physics of Financial Services |

125 Capital markets Sector exploration

126 | Sector Capital markets overview and lukewarm growth trading volumes Strict regulations, flat global have hurt capital profitability for most actors in the ecosystem arkets m O v e r v i e w I s s u e s f a c i n g t h e s e c t o r s cope chapter and description Sector 26% fall in fixed - income trading a revenues Falling revenue cross across the top − 2 12 investment banks in the five years leading to 2016 ore services c of assorted financial securities (debt, Capital markets are composed of the buyers and sellers as well as the intermediaries that facilitate these transactions derivatives etc.), equities, . trade This chapter will focus on active parties in transactions and the deal - making process. Post - such as clearing and functions, chapter. are covered in the Market i nfrastructure settlement, High cost of 9.79 billion in fines investment $ − ten paid by the largest 3 Processes participants Market months eight non r traditional 2016 of banks globally in the first - isk Debt and Derivatives market Deal - making p rocess arket m equity and analytics Research • Corporate i ssuers • • issuers Corporate Chapter scope trade analytics and - Pre • banks) (investment anks b (investment ) services owners Asset • Primary • Sales and trading (asset managers ) • Risk and finance support − reliance the global derivatives reporting markets About 80% of on Increasing • Buyers and sellers • Buyers and sellers 4 is handled by DTCC’s Global Trade Repository systematically i mportant Clearing and settlement Covered in Market f (e.g f unds , (e.g . hedge , unds . hedge providers infrastructure commercial anks b f unds pension ) ) are managed by the three − 70% of all global ETF assets • Execution and post - and OTC chapter nvestment I • anks b nvestment I • b anks 5 Secondary largest ETF providers trading services trends Sector stemming from Brexit, rising interest → rates and moderate yet volatile Macroeconomic uncertainty growth are dampening trading volume on a global scale Difficulty − goes to 12% of global private capital investment accessing New such as regulations are decoupling → the Markets in Financial Instruments Directive II pportunities merging o emerging markets, despite the fact that they account in e for banking services and research fees 6 most of markets global GDP growth fixed Declining - → income trading revenue is slowing down industry revenue growth in the short 1 term → I ncreasing maturity and rising complexity of Asian emerging economies are creating and other apital c that are becoming integrated into the new markets and institutions ecosystem arkets m in gross − - year low Ten market value of outstanding over - Pressure on d erivatives - → formation - capital assets (e.g. - Crypto are allowing non ICOs) traditional players to introduce new 7 edging markets 2017 and h the - counter (OTC) derivatives as of July the future and regulatory treatment of which is unclear methods, Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 125 The New Physics of Financial Services |

127 ummary | Strategies s markets Capital AI has the potential to democratize access to capital across the global economy safety capital m arkets , efficiency greater by unlocking and performance in es in The rise of AI will initiate and accelerate the following chang capital m arkets : Capital and risk management will Deal - Financial models will become Specialized digital tools will making and investor accurate more - matching will be increasingly proliferate across the deal , as firms , improving the ability be a differentiator , reducing the defensible financing to to extend automated early specialize in new margin - stage or making value chain , shifting the distressed companies connected” - “best advantage of the and offsetting tools calculations workload of bankers in both the front back office firms and requiring new core and the competencies include: arkets m capital enabled strategies in - Key AI Strategy C Strategy B Strategy A p erformance Improve investment advanced c Deploy apital - risk and - deal Simplify the making p rocess management solutions through r esearch insights C B A E.g → u se machine learning to detect predictive analytics to se u . . E.g → machine learning to pinpoint se u . E.g → deal prospects based on non - - testing automate due diligence and report anomalies in stress traditional or unstructured datasets results key examples the components and detail The following slides will explore each strategy, highlight Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 126 The New Physics of Financial Services |

128 - making | Strategy A: Simplify the deal markets Capital p rocess A AI can perform administrative tasks faster and better than humans, value activities - enabling the latter to focus on higher W h y s i m p l i f y i n g p r o c e s s e s m a t t e r s Scope of r falling Prediction of evenues ffering xperiences e responsive and ast f for Need o market Fees in the global IPO market have declined from $12 As commercial - diligence tools become more Lean operations allow for opportunities to service smaller due 8 pushing institutions deals and clients, billion in 2007 to $4.3 billion in 2017, improving the accessibility of otherwise important in the digital environment, firms will need to cost capital market offerings enable faster high - decision - to look to cost savings for profit growth making on possible deals O b s e r v e d o p p o r t u n i t i e s Automate pre and administration diligence trade analysis, due - Automate investment monitoring and reporting oint Pain p oint Pain p to receive information in near - real Investors want - time on investment and deal performance to - deal analysis, which There are a large number of manual and low - value tasks throughout pre investment professionals highly trained are often completed by track how key financial metrics are developing pricing Due diligence, prospectus preparation, roadshows, - intensive labour Most communications with clients require report preparation and data extraction from different sources. and other processes are and slow, yet they must be handled by This slows client service as highly staff. Many rote tasks in these processes, such as skilled the information will not always be ready for communication immediately, can be automated, freeing up banker capacity to and manual work is required to source data and prepare reports. This process can stretch a single analysing documentation and legal requirements, - value interactions update request to focus on higher many days New c New c apabilities apabilities Digitized information sharing enables preparation tools and  Automated investor reports can streamline  New  Voice assistants allow clients to ask an document -  and receive institution specific questions content such as reports, performance data, scale automation - platforms allow for the large the valuation process at scale through the 24/7/365, reducing the pitch collection, normalization and analysis of decks and legal documents to be easily personalized answers of key services such as due - diligence financial data using document recognition circulated with deal prospects in open and documentation and routine preparatory tasks reliance on manual work for rote question and answer automated virtual data rooms AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples Sachs’ due platform, CENTRL automates data aggregation for Goldman diligence Deal Link digital interface has ‘s automated Clearwater - Deutsche Bank and IBM Watson have to automate half of the 127 steps of investment accounting and reporting by a partnership to implement a Assess360, provides a central repository for all enabled it announced 9 checklists diligence content, where all parties can view risk automating data intake and making it easier to cognitive advisory tool to support clients and informal IPO 10 processes detailed analytics profiling and in internal and external produce necessary reports instantly advisers Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 127 The New Physics of Financial Services |

129 p t markets | Strategy B: Improve investment Capital erformance nsights hrough r esearch i B tracking down AI can help discover promising investment opportunities by research methods that are not detectable through conventional patterns W h y a d v a n c e d r e s e a r c h i n s i g h t m a t t e r s new Reliance on quantitative buy - side processes Risk of tacit i e ntrants nformation Growth in on relationships to identify deal - rely over Investors often Sell New players are automating relationship - based processes - side institutions struggle to provide services to and mining public data to find companies and teams with quantitative investors, particularly when pitching private opportunities, even through data - driven, analytical unique features companies where data is less available approaches are now available O b s e r v e d o p p o r t u n i t i e s better investment performance by using pairing and sales activities Achieve Use predictive models to improve deal identification, new data in opaque markets Pain p oint Pain p oint - IPO stage) is slow and Market and deal analytics for growing companies (e.g by referrals and labour intensive . Historically, the capital - intermediated, driven has been highly raising process - and inefficient labour face networking, making it intensive - to - face companies, compared to analyse target There is notably less data available for venture capital firms to - later happen in Firms are reliant on, and limited to, the knowledge of brokers, resulting in suboptimal pairing frontier industries where rules of stage organizations. Furthermore, many investments - and investments. This reduces the overall efficiency of capital and trends are often unpredictable. Even for later m arkets and performance, stage companies with predictable financials, scaling investors analytics is challenging when certain information is missing (e.g . audited financials ) both for investors and growing companies apabilities New c apabilities New c  can support  Synthesis of deal prospects and “partners using machine Increased speed of analysis  (e.g . old  Analysis of unstructured data Algorithmic matching systems deals the creation of pitch books by determining the , total money raised, founding using machine learning invest” most likely to learning allows institutions to test correlations - team most important factors that investors consider using machine learning can in a more automated way. This reduces the can improve sales efficiency. Proprietary data background etc.) and synthesizing key points, which will better generate indicators of future success effort required to build models, freeing from investment banks can be used to identify and match deal prospects and deal resources to do more qualitative analysis understand the views and preferred opportunities of different investors . opportunities interviews) (e.g AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples “Emerging introduced the JPMorgan uses a system to find and catalogue Kognetics and others are adding quantitative analysis to HOF Capital, Hone Capital, Correlation Ventures Engine”, which helps identify venture financing by building tools to filter for deal prospects using criteria such as number of PhDs, data, to identify attractive acquisition candidates Opportunities clients best suited for follow founders who worked at successful technology companies, and tier on equity offerings - industry in the technology - one university team members. cost at low analysed Datasets can be pulled from public sources such as Crunchbase and via automated analysis of financial positions, 11 data market conditions and historical Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 128 The New Physics of Financial Services |

130 c risk Capital markets | Strategy C: Deploy advanced s apital and - management olutions C to track their risk exposure more accurately and allows institutions AI optimize capital reserves in real time b e t t e r r i s k m a n a g e m e n t m a t t e r s W h y osts Pressure from new r Increasing margin egulatory r anagement m isk r c ccurate a more Need for equirements , margins There is mounting pressure to increase capital Increased customization of risk profiles allows for earlier Dealers have changed the way they price OTC derivatives optimization capital requiring more sophisticated and more accurate risk estimation, which can limit . to reflect funding costs (e.g - funding initial margins for techniques centrally cleared derivatives) non - unwanted exposure O b s e r v e d o p p o r t u n i t i e s - - time pre - and post Develop real trade risk management solutions Use broader and better data to develop predictive risk models that drive capital savings oint Pain p oint Pain p - ties up capital and increases costs, particularly when risk due to requirements Excess capital must be kept to meet liquidity model Inefficient calculation of initial margins inaccuracies from leveraging - de regulatory frameworks are increasingly focused on risk mitigation and limited use of data processes are a function of several optimization - margin Initial risk measurement is highly dependent on the quality of data used by risk models, as well as such as counterparties and Accurate variables, underlying assets. Standard optimization algorithms are limited in their ability to take complex mixture the risk methodology itself. A limited number of datasets are currently used compared to the wealth of a and semi of parameters into account without significant unstructured datasets available, which could structured - effort help validate risk models apabilities New c apabilities New c Pre -  Continuous risk modelling  Analysis of new risk parameters that enables  Scrutinizing alternative datasets with trade risk analysis can determine the  machine impact of different trade scenarios on overall institutions to automate risk models, influence initial margin requirements via - algorithms can improve the learning coverage and granularity of risk models, as offsetting pairs real time changes to exposure in understand machine learning (e.g . portfolio positions and factor in the cost of risk of and trades, offsetting strategies at the same capital in profitability calculations well as improve the quality of data fed into the recalibrate capital levels dealer, and innovation in trades from one system overall dealer to another) AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples equity derivatives JPMorgan DTCC is testing services that inform clients of has redesigned its Standard Chartered published Natixis’s Kinetica provides investment firms with tools for - real near time tracking of risk exposure, the capital Market business uses machine learning - impact of trading strategies Risk platform, which now - ratio where AI cases paper outlining a to detect anomalous projections enabling the continuous monitoring of capital algorithms completed after, before, rather than trades are manages over one billion risk outperforms standard for optimizing margin requirements testing - sensitivities and provides generated by its stress valuation 12 14 13 adjustment visibility 17 times faster models Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 129 The New Physics of Financial Services |

131 Capital markets | Conclusion Looking forward - While increased efficiency in deal the global economy, supervisory bodies making may have clear benefits for 1 will struggle to keep pace with innovation, creating potential gaps in macro - prudential oversight - - exchanges and over in both listed AI will be core to the matching of buyers and sellers, the counter markets 2 As portfolio management is increasingly side institutions - and customized to the individual, sell digitalized 3 may consider the opportunity to engage investors and consumers directly without intermediaries Market participants must ensure an ethical, transparent and explainable distribution of capital, which may 4 require new tools and processes when these decisions are aided by AI Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 130 The New Physics of Financial Services |

132 Capital | References markets References revenue “Wall in Free Fall, and Here's Why”, as of 19 April 2016. CNBC . Retrieved from https :// www.cnbc.com/2016/04/19/wall - street - banking - Banking Revenue Is - is - in - free - fall - 1. Street - heres - why.html and “What’s $27 Billion to Wall Street? An Alarming Drop in Revenue”. The New York Times . Retrieved 2. https :// www.nytimes.com/2018/01/11/business/wall - street - goldman - sachs - from fixed income - bond - trading.html - - “Investment Financial Times . Retrieved 3. https :// www.ft.com/content/83dfde22 Bank Fines on Course to Rebound”. 7115 - 11e6 - a0c9 - 1365ce54b926 from 4. “ Policymakers Left with Problem in the Wake of London W hale”. Financial Times . Retrieved from https :// www.ft.com/content/7b15e4c2 - 638c - 11e5 - 9846 - de406ccb37f2 5. Three Largest Players Have A 70% Market Share In $4 Trillion Global ETF Industry”, as of 17 May 2017. Forbes . Retrieved from “The https ://www.forbes.com/sites/greatspeculations/2017/05/17/the - three - largest - players - have - a - 70 - market - share - in - 4 - trillion - global - etf - i ndustry/# 15498ead61f6 - 6. itfalls for Private Equity Firms”, as of August 2016. INSEAD. https :// centres.insead.edu/global P private - equity - initiative/research - “Private Equity in Emerging Markets Changes and publications/documents/IE2 - PEinEM - PitfallsandChancesforPEfirms - RaphaelVantroost - FINALSep16.pdf 7. “Statistical Release: OTC Derivatives Statistics at end June 2017”, as of November 2017. Bank for International Settlement s. Retrieved from https://www.bis.org/publ/otc_hy1711.pdf - 8. “Goldman’s Rise of the IPO Machines”, as of 14 June 2017. Bloomberg . Retrieved from https :// www.bloomberg.com/gadfly/articles/2017 robotic 06 - 14/goldman - sachs - s - - ipos - address - a - shrinking - - market equity "As Goldman Embraces Automation, Even the Masters of the Universe Are MIT Technology Review . Retrieved from 9. Threatened”, as of 7 February 2017. - - - embraces - automation - even - ://www.technologyreview.com/s/603431/as - masters goldman of - the - universe - are - threatened / https the “Deutsche Bank Jumps 10. Ambitions”, as of 12 October 2017. Banking Technology . Retrieved from https :// www.bankingtech.com/2017/10/deutsche - bank - on IBM Watson for Germanic AI jumps on - ibm - watson - for - germanic - - - ambitions/ ai 11. “Machine Learning Is Now Used in Wall Street D ealmaking , and Bankers Should Probably Be Worried”, as of 4 April 2017. Business Insider . Retrieved from http :// www.businessinsider.com/jpmorgan - using - machine - learning - in - investment - banking - 2017 - 4 www.businessinsider.com/jpmorgan 12. s Targeting Silicon Valley”, as of 4 April 2017. Business Insider . Retrieved from http :// I - annual - report - on - tech - 2017 - 4 “JP Morgan 13. “Evolutionary Algos for O ptimising MVA”. Risk.net . Retrieved from https :// www.risk.net/cutting - edge/banking/5374321/evolutionary - algos - for - optimising - mva managers 14. Risk.net. https :// www.risk.net/risk - management/4646956/model - risk - “Model Risk Managers Eye Benefits of Machine Learning.” - eye - benefits - of - machine - learning Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra . Wild - 131 The New Physics of Financial Services |

133 Market infrastructure Sector exploration

134 | Sector overview infrastructure Market The shift of service offerings beyond trade facilitation exposes infrastructure providers to new strategic and operational challenges O v e r v i e w I s s u e s f a c i n g t h e s e c t o r s chapter and description Sector cope 6% growth a cross − annual revenue gains for market infrastructure Uneven 3% g eographies in EMEA 1% and US in the Pacific vs. - providers in Asia 3 in 2017 Market infrastructure refers to all entities providing venues, information, networking and IT, custody, services to . and settlement services to capital market participants engaged in trading (i.e clearing buy broker dealers, and sell ). actors - - side - side and fintechs ntrants New e − and was invested in 15% of global fintech venture capital This chapter will focus on underlying infrastructure products and service providers, while active ervices s ntering exchange e capital market infrastructure fintechs in 2016, up from less financial market participants are discussed in the Capital m section of this report. arkets 4 than 1% in 2011 Covered in Capital Chapter scope − 61% of World Federation of Exchanges members markets believe start companies will have ups and large technology - the most impact on market infrastructure in the next five Core Post - technology trading - and trading - Pre making Deal Information and 4 p execution rocess p i nfrastructure rocessing data years market • Primary • Clearing Order • • Data a - sset classes lag in Just 19% of the trading volume of investment grade − C ertain ata reporting d management • Custody management moving to electronic corporate cash bonds is conducted electronically in the US Surveillance of • Settlement • Network and IT • Price discovery • 5 activities • Market making compared to around 90% for equities trading 2018, as of early • Risk 6 and spots forex management competition 45% of Intercontinental Exchange’s revenue was Increasing − trends Sector p nformation i roviders by data services as of 2017, up from 22% ith generated w 7,8 in 2014 as more countries impose central clearing Increasing use of central counterparty clearing → obligations and gradually extend them to new types of derivatives and counterparties − trading are handled by 95% of OTC interest rates derivatives Concentration of → of exchanges in spite of rising political and regulatory appetite for consolidation Continued 9 1,2 LCH.Clearnet c ritical providers a mong border deals - hurdles to cross − clearing is conducted by 98% of credit default swap as infrastructure providers have aggressively Increasing competition in data markets → 9 Intercontinental Exchange diversified beyond trading activities and into data and analytics products initial - margin and − 88% of financial resources , including → of data hosting and non core functions to technology service providers, - Gradual outsourcing default funds, sit in ten central counterparty clearing based on rising confidence in the security and operational benefits offered by cloud operators 10 houses (CCPs), raising concerns from regulators Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra. Wild - 133 The New Physics of Financial Services |

135 Market infrastructure | Strategies summary can bolster the resilience AI and efficiency of market infrastructure while allowing providers to augment their value proposition through new services changes The rise of AI will initiate and accelerate the following in market i nfrastructure : Supervision of the financial New asset classes will move to Data services will be a core part Speed of transactions for most to of the infrastructure provider is likely as AI overcomes electronic trading asset classes as system will be reshaped and - as automated business model through management - straight as accelerate - risk sharing and data detection - fraud advanced processes enable security features limit the ability of quality - high challenges for illiquid securities processing increases the speed of settlement content production malicious actors to act without that is both faster lower cost and detection or retaliation nfrastructure Key AI - enabled strategies in market i include: Strategy A Strategy D Strategy C Strategy B and risk and new settlement added Offer advanced c - d value Develop ata ompliance Introduce and rder o rocesses p trade - post Streamline and nalytics services a types management “as a service” c increase fficiency e ost D A C B u . se → AI to compute economic image recognition and → E.g . u se → machine learning to develop u predictive analytics to se u . E.g se E.g → E.g . robotic process automation to indicators and swiftly compile optimize order execution in unstable services that track down potentially market conditions marketable trading data and reports reconcile trade data fraudulent trading activity and filter out false positive flags highlight The following slides will explore each strategy, detail the components and key examples Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra. Wild - 134 The New Physics of Financial Services |

136 p and infrastructure | Strategy A: Streamline post - trade Market rocesses increase c ost e fficiency A AI allows institutions to automate reporting and better integrate workflows, through processing reducing manual labour and improving straight - W h y s t r e a m l i n i n g m a t t e r s automated and settlement Increasing risk awareness Speed of rocessing p regulations Support for Clients are increasingly demanding faster, more Firms today are more open to automated Increasing focus on risk management is motivating market trade - post connected solutions that were previously deemed too sensitive to participants to seek solutions that address inefficiencies and more tailored functionality from market in infrastructure providers trade processing post operational risk to change - O b s e r v e d o p p o r t u n i t i e s - through processing - trade workflows to achieve straight Automate reconciliation and incident reporting to improve service quality and cut costs Integrate post oint Pain p oint Pain p Post systems, requiring Reporting and reconciliation of irregular transactions often requires manual investigation, trade processes are slow and costly due to the amount of disparate - regardless of the severity of the abnormality manual operation and input - - trade processes involve several disconnected trading and back - office systems as well as external trade processes are one of the largest sources of errors in both trading and operational risk Post Post management due through processes. The fragmented - and vendors, which limits straight institutions interfaces across cycle - life to the level of manual intervention required in processes such as trade nature of capital markets creates many breakage points where systems are not management, position reconciliation, trade fails management and reporting interoperable New c apabilities New c apabilities allows and reconciliation of  Data normalization can detect input and review Automation of  using machine   Robotic a utomation Analysis of process failures process type, facilitating interoperability institutions to automate systems integration across learning allows firms to search for patterns in trades can reduce manual efforts in workflows output different systems with different data incident data to predict future incidents and and optimize cycle times, generating greater with external workflows with limited prepare advanced incident and exception headers and reporting frequencies efficiencies and minimizing operational costs standards, reprogramming, simplifying the development of automated multi - vendor platforms management AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples truePTS Depository Trust & and between Euroclear , a joint venture Euroclear Global Collateral - DTCC the , owned by the exchange trueEX, is a post - platform using robotic process processing - trade and other automation techniques to automate manual post NLP automation, - trade processes. It offers Clearing Corporation (DTCC), and NEX, announced a new partnership to streamline and improve OTC call the entire deal - capture technology process on the triResolve matching and validation engines as well as AI voice - margin derivatives call processes. This manages - margin 11 processing Margin platform, achieving greater straight - through Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra. Wild - 135 The New Physics of Financial Services |

137 infrastructure ompliance risk management ‘as a service’ and Market c | Strategy B: Offer advanced B AI is creating new opportunities to develop software ‘as a service’ solutions clients pressures that address the regulatory and compliance faced by W h y e f f i c i e n t c o m p l i a n c e m a t t e r s equirements r regulatory Increasing regulatory changing - Constraints of p traditional s ervice o fferings ressure Ever programmes product comply with new continuously working to Institutions are providers’ “core” New regulatory shelves are are increasing reporting Infrastructure requirements, creating the opportunity to market , creating the opportunity for internal capabilities to limited regulations. Technology services have the opportunity to standardized and scalable solutions generate new revenue streams centralize this effort O b s e r v e d o p p o r t u n i t i e s Deploy holistic market - surveillance services Automate alert triage, investigation and reporting Pain p Pain p oint oint A significant number of compliance costs arise due to regulatory reporting requirements for Market surveillance is performed inefficiently, with work duplicated across many institutions - Rules transactions that have been flagged as suspicious based approaches to detect market abuse are based on retrospective flagging of transactions using individual institutions’ incomplete transaction histories. While every financial institution needs to Every positive alert in today’s and classified engines needs to be validated, surveillance - market insights from cannot use adhere to certain surveillance requirements, systems that are built internally documented. This is often a manual process that requires the collation of data from various sources in systems transaction patterns across institutions, limiting the accuracy of internal order to complete paper reports that will be submitted to regulatory bodies and risk groups apabilities New c New c apabilities with based on severity and potential Advanced surveillance systems using    Automated report drafting can allow d raft Modularized and digital systems  Score alerts exposure of the client or the institution to risk. straight - through processing can be deployed machine learning can bring together reports to be issued upon signal of an alert, order Institutions can use this scoring to surveillance “as a service” for - market” as market providers’ “whole ensure and reducing the processing time from when each data other unstructured data ( clients, reducing their infrastructure and e.g analysts prioritize the most pressing . traders’ alert is generated through to completion of a investigations implementation requirements while increasing report and action, if required messages) to increase accuracy and reduce surveillance accuracy rates false positive AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples decision analytics to Next’s surveillance system - market Software AG’s - B tools flag support - uses speech time - real Sybenetix’s - “360 to deliver AI uses Nasdaq’s SMARTS CMC:Suite side and historical data across degree surveillance” for suspicious market activity to compliance search and analyse trading teams - call recordings buy - time - analyses real and sell - voice capabilities asset classes and markets to create flags for positives analytics to detect false and provide and expedite investigation clients, covering significantly higher global 12,13 market standards patterns positive and negative trading Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra. Wild - 136 The New Physics of Financial Services |

138 - Market s ervices and ata d added value Develop | Strategy C: infrastructure analytics C insights by AI is allowing infrastructure providers to introduce new experimenting with data their unique access to trade W h y n e w s e r v i c e s m a t t e r Excludability of Value arket m wnership - o chain diversification data Emergence of a new data Providers combined with new in multiple markets across asset the provision Potential new revenue streams exist through that operate data, Increasing demand for trade insights and will lead to accelerated product of multifaceted datasets , tools for pre - technologies, classes are in a unique position to create valuable and development and sharpened competitiveness and insights predictions, and other services proprietary datasets O b s e r v e d o p p o r t u n i t i e s Develop macroeconomic indicators using internal data and risk structure Offer insights on market “as a service” Sell internal analytics capabilities oint Pain p Pain p oint Pain p oint Market participants use imperfect indicators for valuation Customized insights solutions that use internal and external Market microstructure presents risks and costs that often go data are difficult to build unaccounted Macroeconomic indicators are often presented as periodic goes unanalysed generation insight services will need to there is a delay in as indicators based on surveys. This means Data necessary for - Information contained in order books often next models, - pricing and order data) market data (e.g institutions lack the access, expertise or resources to effectively with non market data into shifts in macroeconomic factoring rapid combine . participants traditional external data sources (e.g . news analysis) alongside creating a risk of mispricing for capital markets understand the data tools expertise, and AI apabilities New c apabilities New c apabilities New c and the learning, using machine Macroeconomic forecasting can be calculated time transaction cost and analytics Real  Advanced analytics engines  Flexible data integration  -  using machine learning to understand how certain trades or depth of price and order data within infrastructure providers, and proprietary tools capable unique and can generate can allow of interpreting raw data can analytics services that can liquidity, trading strategies will affect a variety of indicators (e.g customized insights the provision of new . using - time real internal data, third - party data . high offer unique insight into the frequency traders) predict market performance in bid/ask spreads). Participants (e.g and data from clients behaviour of different classes could use these analytics to plan their trading strategies sources of traders Examples AI and/or Advanced Automation AI and/or Advanced Automation Examples AI and/or Advanced Automation Examples - S&P Global has acquired Kensho, Deutsche Börse is expanding its data service to provide high the Intercontinental brings together data from ICE Data Services - machine a provider of , tools Exchange, New York Stock Exchange and the International Data learning presence its allowing S&P to deepen and analytics raw data from across the order with low latency, life cycle volume 15,16 chains value customer in and Corporation that can noise and rates error be used in economic analytics, - deal and connect with clients during the valuation 17 14 discovery and pre - trade analysis processes solutions connectivity Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra. Wild - 137 The New Physics of Financial Services |

139 | Strategy D: ypes t settlement and rder o new Introduce Market infrastructure D AI allows institutions to deploy new order types settlement methods and - investors term and risk averse - that protect long W h y b e t t e r e x e c u t i o n m a t t e r s Better segmentation of customers history data - Competing on b order Unique est execution Long Infrastructure providers’ unique access to full trade The proliferation of complex and opaque trading practices - term investors have different requirements and risk the speed and across traditional public exchanges has increased histories means they are best positioned to model the appetites to arbitrageurs when it comes to impact of order types and design new products price of execution demand for transparency and fairness O b s e r v e d o p p o r t u n i t i e s term investors - models for long protection - trade Deploy Enable dynamic order types Use dynamic execution to improve trade speed and price Pain p oint oint Pain p oint Pain p strategy options for investors affect Signals created by orders can adversely Limited order types restrict are exposed to adverse order outcomes Investors execution price - surveillance activities are retroactive rather than proactive, Market design can mean the Institutional clients create market signals when they undertake and Market outcomes of transactions the market. By limiting order averse investors to negative effects from illegal portfolio or sector rotation, or make changes that require large exposing risk - auctions have a significant effect on . market transactions. These signals can be used by traders to profit activities, predatory trading strategies types to a simple set, market designers are actively restricting the or price quote instability (e.g goals frequency can seek from trading) - that investors from a changing order book, resulting in worse prices for high institutional clients and often slower and more uncertain execution New c apabilities apabilities New c apabilities New c execute trades in to can be used engines optimization - Trade  using machine models surveillance using Trade fragmentation  impact - time price - Real   Predictive market - modelling of optimizing from predatory, high - machine apply can impacts AI can market auctions with the goal learning can anticipate adverse certain metrics ( e.g. large order into divide a smaller frequency trading strategies and protect orders from retail or learning to predict the price volume, price or speed) placed over a orders longer time horizon to best impact of a given order, and term investors - long resulting the mitigate price movements and transaction costs execute at the best price AI and/or Advanced Automation Examples Examples AI and/or Advanced Automation AI and/or Advanced Automation Examples advertisement uses machine learning in its real learning IEX techniques to create JPMorgan is testing a system called LOXM in its European has introduced machine Google Trading - time - - placement marketplace. allows companies to place ad orders his new order types how to execute client orders with business. It learns equities that protect trades from T during unstable, execution . best bid or best offer is about to maximum speed at the best price ( how best to offload big on the Google AdWords network based on additional objectives e.g. potentially adverse conditions (e.g 20 18 19 decline or increase) conversion) as opposed to only (e.g prices) price equity stakes without moving market . to maximize Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra. Wild - 138 The New Physics of Financial Services |

140 Conclusion Market infrastructure | Looking forward N ew service offerings will make infrastructure providers more resilient and improve participants’ access to 1 questions about ownership of data and conflicts of interest analytics, yet they will also raise participants risk being Increased automation will make global trading much more streamlined. However, 2 further malicious activity from new participants integrated into digital trading networks to exposed i market becomes increasingly nfrastructure global, digital and complex, institutions will struggle to track As 3 requirements their fragmented regulatory become markets, new standards may be critical to the functioning of the financial As data - service providers 4 necessary to ensure the quality , accuracy and availability of data I ncreasing the speed of transactions and the velocity of money will challenge institutions and regulators to 5 address the increased risk of financial contagion during market turbulence Card Scenarios Sector Explorations Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra. Wild - 139 The New Physics of Financial Services |

141 Market infrastructure | References References - of Global Exchange M&A? Not It”, as of 28 March 2018. Bloomberg. Retrieved from https :// www.bloomberg.com/news/articles/2018 Duffy Can Help 03 - 28/death - of - “Death 1. if CME’s exchange - m - a - not - if - cme - s - duffy - can - help - it global - Financial Times “Nasdaq CEO Says Era of Exchange Consolidation Is Over”. https:// www.ft.com/content/9d60b1d4 - 27b0 - 11e8 - b27e - cc62a39d57a0 2. . Retrieved from McKinsey&Company from “Capital Markets Infrastructure: An Industry Reinventing Itself”. 3. . Retrieved www.mckinsey.com/~/media/McKinsey/Industries/Financial%20Services/Our%20Insights/How%20the%20capital%20markets%20infrastructu re% 20industry%20is%20reinventing https :// industry the capital - markets - infrastructure - - - is - reinventing - itself.ashx %20itself/How - “Capturing the Opportunity in Capital Markets Infrastructure”. World Federation of Exchanges . Retrieved from https ://www.world - exchanges.org/home/index.php/files/18/Studies --- 4. --- - - Fintech Reports/497/WFE Decoded - report.pdf McKinsey: “MiFID II Poised to Increase Electronic Trading, Tighten Spreads in European Fixed Income Market”, as of 15 March 2018. Greenwich Associates . Retrieved from 5. - trading www.greenwich.com/press release/mifid - ii - poised - :// - electronic - - - tighten - spreads - european - fixed - income https increase “Electronic Trading in Fixed Income Markets, as of January 2016.” Markets Committee. https://www.bis.org/publ/mktc07.pdf 6. “Delivering Vital Infrastructure and Information to Transform Global Markets”. from . Retrieved 7. https :// www.theice.com/publicdocs/ICE_at_a_glance.pdf Theice “Intercontinental Exchange Fourth Quarter & Full Year 2014 Earnings Presentation”, as of 5 February 2015. Retrieved from 8. Intercontinental Exchange. ://ir.theice.com/~/ media/Files/I/Ice - IR/events - presentations/presentation/4q14 - earnings - presentation - v1.pdf https if 9. Clearinghouse Fails?”, as of 6 June 2017. Brookings . Retrieved from https ://www.brookings.edu/research/what - if a - a - clearinghouse - fails / “What 10. “CCP Study Raises Concentration Concerns”, as of 5 July 2017. Reuters. Retrieved from https :// www.reuters.com/article/ccp - study - raises - concentration - concerns - idUSL8N1JW2ZA 11. “DTCC - EUROCLEAR Global Collateral and NEX Partner to Further Streamline Margin Call Processes”, as of 11 April 2018. DTCC . Retrieved from http :// - euroclear - global - collateral - and - nex - partner - to - further - streamline - margin - call - processes www.dtcc.com/news/2018/april/11/dtcc “SMARTS - Sell - Side”. Nasdaq . Retrieved from http :// business.nasdaq.com/market - tech/market - participants/SMARTS - trade - surveillance 12. sell - side Trade Surveillance System for the “Best Market Surveillance Product: www.risk.net/awards/5295116/best SMARTS”. Risk.net. Retrieved from https :// 13. - market - surveillance - product - nasdaq - smarts Nasdaq data 14. Services”. Theice . Retrieved from https :// www.theice.com/market - Data “ICE 15. “Future of Fintech in Capital Markets”, as of 20 June 2016. Deutsche Börse Group. Retrieved from http :// deutsche - boerse.com/blob/2621702/ed055219caeb553f43950609d29e1bb3/data/future - of - fintech - in - capital - markets_en.pdf us/news 16. Data for the Financial Industry”, as of 5 March 2018. Deutsche Börse Group . Retrieved from http :// www.mds.deutsche - boerse.com/mds - en/about - of - highlights/the - “The Value value of - data - for - the - financial - industry/14822 - 17. “S&P Global to Acquire Kensho; Bolsters Core Capabilities in Artificial Intelligence, Natural Language Processing and Data Analytics”, as of 6 March 2018. CISION PR Newswire . in Retrieved https://www.prnewswire.com/news - releases/sp - global - to - acquire - kensho - bolsters - core - capabilities - from - artificial - intelligence - natur al - language - processing - and - data - analytics - 300609450.html Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra. Wild - 140 The New Physics of Financial Services |

142 Market Infrastructure | References References 18. “5 Things You Should Know About IEX”, as of 16 June 2016. The Wall Street Journal . Retrieved from https://blogs.wsj.com/briefly/2016/06/16/5 - things - you - should - know - about - iex/ https://www.thinkwithgoogle.com/intl/en 19. “Adwords Smart Bidding Playbook”, as of May 2017. Google. Retrieved from bidding - gb/advertising - channels/search/adwords - smart - / www.ft.com/content/16b8ffb6 20. “JP Morgan Develops Robot to Execute Trades”. Times . Retrieved from https :// Financial - 7161 - 11e7 - aca6 - c6bd07df1a3c Sector Explorations Card Scenarios Key Findings Cross - Sector Impact Dep. & lend Insurance Payments Inv. mgmt. Capt. mkts . Mkt. infra. Wild - 141 The New Physics of Financial Services |

143 How might the continued evolution of AI the future affect of financial services? Selected “what if?” financial services for the long - term implications of AI for scenarios

144 Future scenarios | Overview new leaps in capabilities, causing AI The continued evolution of will enable financial services disruptions further in cenarios driven uture f Overview of AI - s currently available and AI’s development is unpredictable and, through our research, we encountered several future scenarios that may not be feasible given the technology fundamentally different ways of thinking the existing regulatory structures. Nonetheless, these scenarios represent about financial services that merit consideration in the public and private sectors alike Selected future s cenarios Insurers “lines as act c tility raud f laims A c entral u Disappearance of Death of deposit Rise of a master of d efence ” is formed funds a ccounts m nsurance i odel - A Everything is managed w ealth assets are central anti - money Illiquid laundering t Online platforms b ecome d ebt okenized formed u tility w arehouses is these scenarios across sectors, examining implications as well as degree of possibility their The following slides will explore - Card Scenarios Key Findings Cross - Sector Impact Sector Explorations Wild 143 The New Physics of Financial Services |

145 ccounts a deposit Future scenarios | Death of was continuously and automatically allocated across What if ... cash financial products, leading to the death of deposit accounts? What would be the implications? What if...? where money sits accounts, to exist: earning no interest, idle Common deposit cease savings through optimized improved Customers’ financial health is radically centre Wealth managers move to become the of the retail customer experience, managing all flows of • rates and automated loan repayments that minimize interest expenses money for customers • F unds flow to and spending categories debt, savings – matching the “optimal” account, including Deposit accounts are no longer the locus of control for customers as the of their overall budgets and accounts customers’ “mental accounting” management platforms, financial centre of the retail customer experience shifts to - reducing interaction points between large banks and customers Manual decisions required Reducing debt Demand deposits shift further into the capital markets and wealth and industry w Savings and ealth Current away from the balance sheets of retail banks anagement m state Deposit Item Deposit The cost of funds for as liquid deposits become significantly lenders increases e - to - xpenses Day day ) account (e.g . pay cheque increasingly scarce The and wealth management borrowing between spending, demarcation Automatic decisions allow for the flow of customers’ money - as one service down breaks based offering manages Reducing debt user uninterrupted flow of funds to end across multiple accounts Future and Savings nvestments i Current regulatory frameworks, largely focused on liquidity and leverage, state Account deposit on demand centred that is not for a new banking model need to be adjusted AI - powered - Day e day xpenses - to ) . pay cheque (e.g deposits w ealth m anager WHY HASN’T THIS HAPPENED? Why might this happen? • understanding of "mental make optimal choices on behalf of customers behavioural and a their money better, to manage can automatically Customers’ desire • Ensuring that an offering requires the development of AI solutions that can consider a broad set of financial and non accounting”, are the drivers behind this development, which allows better personal financial management. - customer contexts comprehensively By clearly articulating spending categories and allocating funds directly to financial information to understand accounts , this trend helps customers realize their aims better than they could through manually managing their own cash flows regulation in their ability to transfer funds across other restricted by Institutions are • . • Increasing levels of automation pay debts to allow (e.g real time in institutions to optimize deposits transaction customer consent, limiting institutions and across accounts without explicit, per - cut interest expenses), reducing the need for customer interaction the ability to automate flows of money - Card Scenarios Key Findings Cross - Sector Impact Sector Explorations Wild 144 The New Physics of Financial Services |

146 as Future scenarios | Insurers act ‘lines of defence’ What ... in insurers transition to becoming providers of ‘lines of defence’ if resort’? of last ‘funders addition to What if...? What would be the implications? resort, by: in addition to pooling risk as a last risk, Insurance becomes focused on preventing focus on prevention at as the new fewer property losses Customers experience • Using internal and external datasets (e.g . real - time sensor data), allowing insurers to predict loss events insurers reduces their risk exposure - services avoidance deploy loss emptively - pre with greater accuracy and IoT devices, weather • with actionable recommendations based on real time data (e.g - Prompting users . clients are not covered by their current insurance policy and “nudging” them to feeds), detecting when transparency with regard to Policyholders have increased policies and insurance purchase additional and alerted to as coverage exposed on the risks to which they are they are updated situations for which they are not covered Insurer – lines of defence Insurers focus in addition to customers, on creating positive touchpoints with minimizing negative touchpoints that are associated with claims and damages prevention and Risk advisory Future you left your window open in the rain” e.g . “Alert : prices, as reduced loss rates lead to declining models emerge New monetization state and revenue is instead generated by prevention services a policy ugmentation Proactive season approaching . “Hurricane e.g would you – like to add flood insurance ?” to be products need The ethics of data use and targeted pricing in insurance recovery Claims and revisited as highly accurate models create “risk pools of one” and potentially price . “We’ve e.g deposited $500 in response to water certain segments out of the insurance market basement” we detected in your Why might this happen? Why hasn’t this happened? Current analytical tools are not sophisticated enough to enable the precision and accuracy • and public spaces creates a large cars in homes, The continued proliferation of connected devices • in conjunction with advanced predictive models policies, breadth and depth of data that required to provide customers with valuable preventative advice, or to augment in provide is used to – – time advice to prevent losses . real time proactive, real A ccess to the required data is not yet easily available to insurers - • Modularization and commoditization of common individual general lines leads insurers to seek new • Insurers have historically been unable to demonstrate to customers the benefit of ratio optimization - differentiators and sources of loss connected telematics devices . as these are often seen as punitive features (e.g devices, customers’ best punishing bad drivers) rather than as being in interests Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 145 The New Physics of Financial Services |

147 | A u formed is tility raud f laims c central Future scenarios ... What - fraud were routed through a central insurance claims all if prevention payments were made? utility before What if...? What would be the implications? A central fraud - prevention utility for insurance claims is formed, to which all claims are sent, by: central utility would decrease Improved claims experience for since a customers , which uses an advanced, AI - based • Consolidating claims data across insurers into one central utility false positives and reduce the processing time for investigations rates false positive model to more accurately identify fraud and reduce Filtering out fraudulent claims and approving legitimate claims for policy issuers to pay damages • insurers as the utility more efficiently Reduced cost of handling claims for identifies fraud and removes the need for expensive fraud investigations. Insurers 4 2 3 1 5 lower may pass on these savings to customers through premiums Back and Current forth Customer Loss Customer files Verifications are Insurer state because the central accurately identify fraud can more utility losses Reduced fraud occurs purchases claim with insurer conducted provides environment and blacklist malicious actors across the entire insurance insurance recovery External sources of data differences in operating efficiency as differentiators Insurers need to find new Time across institutions partially normalizes due to the central claims fraud utility, Content increasing focus on more direct modes of competition rd 3 data party - Future 3 2 1 larger number of the insurance workforce may be displaced as many current A 4 state - different institutions are made redundant claims verification personnel across Central analytics Customer files Loss Customer Regulators become critical to the development and ongoing monitoring of the engine nsurer I purchases claim with occurs provides central utility to ensure the system is fair and to develop safeguards against abuse Utility validates claim and insurance fraud utility recovery refers back to insurer Why might this happen? Why hasn’t this happened? AI solutions are not yet able to fully automate the claims fraud prevention process, which • • Fraud is a systemic issue as malicious actors often target multiple institutions with similar methods. The development of collective intelligence allows institutions to dramatically reduce the financial cost of fraud relies heavily on manual analysis and human judgement • from more pressing, The focus on fraud protection distracts institutional investment and executive focus is not ubiquitous and • Connectivity and sensor data for insured goods (e.g vehicles, homes) . utility would that competitively priorities. A provide institutions with greater certainty collective significant will not be for several years, until IoT adoption matures, inhibiting the immediate availability they are protecting themselves from excess loss of high volumes of reliable primary data Financial institutions are wary of sharing sensitive customer data with third parties • in the current regulatory environment, reducing the willingness to collaborate - Card Scenarios Key Findings Cross - Sector Impact Sector Explorations Wild 146 The New Physics of Financial Services |

148 Future funds scenarios | Disappearance of What a digital wealth manager invented a platform that built ‘ funds of if ... ’ for each of its clients? one What if...? What would be the implications? focused on self - serve clients, invents an algorithm that can invest in the full market, An asset manager, Customers see lower fees and increased customization through the by: akin to an ETF but at an individual level, disintermediation of several layers of asset management investors to directly own portfolio assets, with custom portfolio weights determined by • Allowing individual their investing needs, effectively eliminating intermediary mutual funds, ETFs or other aggregated reducing fees considerably investment vehicles, and thereby Front - managers need to quickly develop new core end wealth - asset need to own and harvest core as they capabilities now management • Allowing investment m anagers to automatically create optimal asset allocations for investors, in a fashion capabilities akin to funds robo dynamic investment - advisers , but by investing directly in market assets. This offers similar to current strategies capable of being customized, funds as opposed to relying on retail Investments as hedge funds, ETFs the disappearance of funds managers face Asset Bundled and other asset managers go out of business and are replaced with a new Investments investments Current form of asset ownership Returns Returns state Wealth minus fees managers Returns Retail investors minus fees increased herd risk as few platforms and a limited set of Regulators see General funds Investment securities an outsized algorithms eventually influence on the market grow to have Investment Future as - robo advisers face a fierce Independent competitive response Returns - driven Algorithm Individual retail state platform investor to established asset managers pursue further vertical integration compensate for the erosion of their legacy ETF business Investment securities Why might this happen? Why hasn’t this happened? management impact processes ( e.g. risk • Independent robo - advisers can disintermediate asset managers from the investment process, reducing • AI is not yet robust enough for sensitive and high - - and capital adequacy) to function without human oversight, nor can it integrate with non the cost of managing portfolios for clients and bolstering their low - fee value proposition electronically traded asset classes an opportunity to act as first movers and Integrated asset managers ( e.g. Vanguard , BlackRock) have • advisers that disrupt robo deploy research and development capabilities - to fall so that bulk trades and individual • Transaction costs at the venue level need trades become comparable with respect to cost need • to evolve or be automated to Legal requirements for filing and disclosing within funds individual - level funds facilitate Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 147 The New Physics of Financial Services |

149 m Future scenarios | Rise of a master i nsurance odel that incorporates claims, ... What a master AI an insurer builds solution if policy issuance, pricing and other activities into a single model? What if...? What would be the implications? Insurers implement AI to integrate pricing, distribution and claims data into a single real - time as prices are not static Increased short term variance in pricing for customers - making engine that executes all decision decision - - : making by making it more difficult to compare policies – and prone to change in real time • Using claims data to inform and optimize underwriting in real time, ensuring that pricing engines incorporate new information immediately and have the most accurate view of risk Developing “profiles of one” in underwriting through dynamic risk assessments rather than static risk • Insurers’ competitive focus moves to developing the - in - class AI data - best tables, creating a self - learning pricing engine rather than one that requires manual updates from actuaries as model accuracy is a greater determinant of profitability science capabilities, than scale of distribution or composition of risk pools Multi - line insurers can more accurately price risk as their access to a wider Current breadth of data makes them better positioned to build more sophisticated and Visibility across processes is limited state accurate models Pricing data and Claims data and Displacement of talent as the need for actuarial skills in underwriting is replaced by ecisions d d ecisions and the need for technical AI talent to build and maintain integrated insurance Policy pricing models Future odel Master insurance m Distribution Claims as - near New ethical questions surrounding discrimination arise perfect risk state market and risk creating “risk pools profiling may price certain individuals out of the of one” Servicing Why might this happen? Why hasn’t this happened? , run using very different systems and datasets • Insurance pricing is inherently a function of the probability and severity of potential claims with a given • Claims and policy applications are usually - complex data policy, yet today pricing and underwriting engines are segregated from claims requiring adjudication functions. - integration efforts to make this functionality available Connecting these datasets in real time gives insurers improved pricing efficiency that leads to improved Risk • - based pricing is heavily regulated; many insurers are not ready to switch to a fully AI - profitability driven model due to uncertainty over performance, auditability and potential for bias - based data - driven solutions are already being used in pricing and claims processes, in addition to AI - AI • • While the cost of building such a system would be expensive, the magnitude of rewards integration solutions. The next step is packaging these in a comprehensive, integrated system investment from improved accuracy are uncertain, thus requiring institutions to make a big undefined rewards with Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 148 The New Physics of Financial Services |

150 Future scenarios | t okenized assets are Illiquid ... if illiquid asset classes were fully digitized using AI and distributed What introducing liquidity and easy trading? ledger technology, What if...? What would be the implications? infrastructure, real estate, private equity/venture capital) were fully . A variety of illiquid assets (e.g of alternative as these asset classes assets A dramatic increase in the liquidity digitized and easily tradeable, as a result of: range of become available to a broader investors A tokenized system supported by AI that validates information about the underlying assets, as well as • managing . optimal and democratized This enables more the exchange of ownership of capital and assets The creation of new connections between capital and assets, allowing new clearing processes, minimizing the need for price discovery, settlement and oversight manual pools of capital to access alternative assets Sources of Alt. assets Intermediaries capital as increased transparency and liquidity leads to crime Reduction in financial Infrastructure detection easier anomaly Current Retail Brokers Salespeople efficiency Increased price and market equity Private lower as increased trading activity leads to state and agents and traders spreads and more accurate pricing (who facilitate (who provide price information flow) estate Real discovery etc.) Institutional cross Increased border complexity since the characteristics of many illiquid - assets vary significantly by region Sources of Capital Alt. Assets vendors, as Disintermediation of firms that act as brokers and information Direct access to alternative assets, facilitated by: more investors gain direct access to alternative asset classes, reducing the need for Infrastructure Future specialized intermediary players Retail equity Private state The need for standards bodies to act as venues for collaboration on the - based price Tokenized information AI Real estate governance of token exchanges exchange discovery engine Institutional Why hasn’t this happened? Why might this happen? • AI can be used to analyse unstructured data to quantitatively evaluate assets (e.g . pricing), reducing the • AI is not yet fully able to models from tacit knowledge to assess the financial metrics build of alternative assets (e.g trust, political risk etc.) information asymmetries that characterize illiquid assets. T his limits the need for specialized players who . aggregators etc.) , (e.g. brokers facilitate the flow of information digital markets are impeded by both scalability and latency, • Existing blockchain - based • Technology is already being used to democratize access to asset classes (e.g . crowdfunded equity), limiting the efficiency that can be delivered by decentralized versus centralized solutions customers are willing to use automated solutions to make investments products in complex suggesting • private and required to enable the tokenization of illiquid assets is often not Information available through digital channels, with limited proxy data available Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 149 The New Physics of Financial Services |

151 anaged Future scenarios | Everything is m w ealth ... all accounts and balances of cash, regardless of where they are What if risk? held, earned a return commensurate with their What would be the implications? What if...? asset managers (e.g. robo ) transform low - value balances into value - advisers enabled Technology - - “dead cash” Fragmentation of holdings and minimization of as customers hold customers by: adding investments for their funds in many interest - bearing accounts, across institutions. T his increases the average return on money for customers Using algorithms that automate • the process of managing money, operating at little or no marginal cost companies across industries, • Striking offering to unlock leftover balances on individual agreements with platforms, margin accounts and excess working capital payments and e - commerce managers lose Wealth customer and asset ownership as wealth management Allowing a cut of the interest income, and nstitutions to earn i • on earnings to their customers passing becomes a subsidiary of the experience offered by other products Investors as navigating a multitude of investment channel - Increasing omni complexity delivery channels becomes a key capability for firms seeking to grow AuM Excess commerce - Payment Pre - E Margin Deposits working payments platforms apps Future accounts apital c is a Scale of as there longer advantage assets is no dominant competitive a value Low - state multitude of sources of investment returns outside asset managers traditional alances b to manage an increasingly complex, fragmented ecosystem Regulators need as the number of investment accounts and investment firms to which an individual is Financial markets tied increases Why might this happen? Why hasn’t this happened? advancements requires Ensuring the suitability of investments across different institutions , enabling and cheaper movement of money faster Modern payments infrastructure allows for • • time data feeds and relate unstandardized datasets institutions to seamlessly move funds across accounts using automated solutions to optimize customer in AI to process real - returns - cost • It is more efficient to manage funds in a single institution, as high transaction costs • individual institutions to build a allows through open banking) . (e.g Expanded access to data level volume deals A . ). with nd - make scale a critical requirement (i.e to receive exchange themselves. comprehensive understanding of customers’ finances, even if they do not hold the funds the current infrastructure, funds cannot flow quickly enough to fully unlock “dead cash” F urthermore , portfolio construction is a relatively well - understood field and increasingly becomes a commodity, allowing for institutions to automate the process Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 150 The New Physics of Financial Services |

152 ecome arehouses w ebt Future scenarios | Online platforms b d commerce and business software platforms become the main - e ... if What point of lending for both retail and commercial borrowers? What if...? What would be the implications? software “as a service” platforms (e.g . Amazon, QuickBooks) enter the lending and B2B commerce - E analytics but as a result of better terms Borrowers are granted highly customized space by: have to entrust more data to fewer, larger organizations Absorbing the adjudication - loan function, allowing them to price and underwrite loans based on their own • proprietary methods Platforms offer deals to encourage customer retention as they can bundle Originating and temporarily warehousing loans on their books, selling them to the highest bidder through • selective credit deals ( margin higher products) with margin - i.e. low - products (e.g . securitized and unsecuritized channels subscription services, retail products) Interest Goods Current and personal data lending profit pools Platforms earn the negotiating majority of by developing state Loan Adjudication Payment favourable the ability to negotiate giving them power over traditional lenders, deals Borrower Loan Lender asset function Retailer with the lowest bidders become balance t sheet providers o customer ownership and Lenders lose Capital markets Interest minus spread platforms, in effect having to compete directly with capital markets participants to investor Retailer secure loan assets Highest bid Loan and goods Future depository - Non ender l Adjudication function state Interest Borrower Regulatory new capabilities to assess risk and to quickly deploy bodies need Bank banks compliance within commercial platforms as well as Loan assets balance s heet Debt warehouse Why hasn’t this happened? Why might this happen? - AI commerce platforms (e.g . Amazon, Alibaba) own the market because they outperform financial E - driven pricing and underwriting algorithms that use alternative data are nascent and • sufficient increased accuracy over traditional methods to warrant institutions when pricing and underwriting loans : have not demonstrated wholesale switching They have greater access to broad datasets, modern infrastructure and more mature AI capabilities, • which cumulatively lead to more accurate and precise lending than traditional underwriting and pricing US under the • Regulations upholding the separation of banking and commerce (e.g . the models to adapt to allow additional lending responsibilities to be ) need Bank Holding Company Act financial commercial entities - As platforms’ relative accuracy increases, financial incumbents reach a tipping point where they earn • absorbed by non higher net interest margins by participating in platforms and paying a spread or fee - Card Scenarios Key Findings Cross - Sector Impact Sector Explorations Wild 151 The New Physics of Financial Services |

153 scenarios tility formed is u laundering - money - central anti | A Future money What if ... anti - - laundering (AML) surveillance were provided to all institutions service provider? by a centralized and collectively owned What if...? What would be the implications? - institution basis, a central utility is formed by: - Instead of managing AML on an institution by as institutions and regulators become more confident inclusion Increased financial prone in servicing customers with limited financial histories, or from more risk - • allow - sharing and analytics models that Consolidated data for new, more robust prevention systems with geographical areas more accurate detection and insight financing Development of • to protect against money laundering, terrorist “collective intelligence” and other systemic threats, particularly threats employing transactions across multiple institutions that would lines Positive impact on financial institutions’ bottom through cost savings in institution otherwise be difficult to detect for any one AML. Spillover benefits include the reduction managing in associated prices and a uplift from targeting untapped potential revenue markets Central AML utility External data sources A consistent as current privacy regimes consent model is required - consumer may restrict the creation of a central utility and cross - border data movement Public Social Co - owned repository party Third - and analytics engine data model is required A redefined liability to clearly identify and define the legal and Shared Shared Shared Results Results Results financial responsibility structure for when AML requirements are breached Future Digital ID Mobile data data data state Increasing concentration of risk as a single false positive or false negative from a through the entire financial ecosystem is propagated central utility as many Risk of displacement for a large number of financial professionals current AML personnel across individual institutions are made redundant Financial institutions Why might this happen? Why hasn’t this happened? • Sharing data to form a central utility would allow institutions to access more advanced capabilities than Current solutions are not sophisticated enough to derive meaningful insights from these • systems would data feeds , which lack standardization and normalization institution even the most sophisticated incumbents could develop independently ; machine - learning fragmented cross - be able to work on a more complete view of the transaction landscape, allowing institutions to identify Financial institutions are wary of sharing sensitive customer data with third parties, due to • suspicious patterns that may be spread across the environment and competitive risks security, regulatory would be able to collectively decrease costs through the commoditized service, removing Institutions • and allowing them to focus investment on efficient compliance as a source of competitive differentiation more critical areas of expertise (e.g. investments in customer experience) Card Scenarios - Key Findings Cross - Sector Impact Sector Explorations Wild 152 The New Physics of Financial Services |

154 Concluding thoughts Next steps for the financial services ecosystem

155 Concluding thoughts AI’s impact on the physics of financial services will demand increased collaboration to address emerging uncertainties AI will introduce several uncertainties to the global financial services ecosystem Systemic safety : As AI creates new types of risk in financial services systems (both at national 1 and international levels), new risk - management and mitigation strategies will be required making processes, new methods of protecting consumers and - : As AI automates decision Consumer protection 2 ensuring the public interest is sheltered will be required force, collective action by Human capital : As AI creates new forms of labour needs and displaces portions of the labour 3 be institutions and regulators will required The World Economic Forum will continue to explore outstanding questions and create venues for collaboration among stakeholder s Bring together diverse groups of stakeholders to explore the potential for collaboration, which can overcome key barriers 1 to unlocking the value of AI Convene industry leaders, regulators and public policy organizations to explore and address emerging societal issues 2 154 The New Physics of Financial Services |

156 Additional reading

157 reading Additional Additional reading learning more detail about the topics exploring further and The following texts were instrumental in shaping the perspectives of the project team. For those interested in covered in this report, we highly recommend reading the following documents: - commerce Social Networks, E The Seven Deadly Sins of Platforms, And The Growth Of Prediction Machines Ghosts in the Machine Predicting the Future of AI Digital Payment Ecosystems In Goldfarb, and Avi Agrawal, Ajay Baker McKenzie China Rodney Brooks Joshua Gans Better Than Cash Alliance When Will AI Exceed Human Artificial Intelligence, Performance? New T Moats he Automation, and the Economy Bank of the F uture Salvatier Grace, John , Allan Katja Jerry Chen and Greylock Executive Office of the President, CITI GPS Baobao Zhang, and Dafoe, Partners Obama White House Evans Owain Intelligence, Ethics and Artificial Artificial Intelligence: Potential Artificial Intelligence And Machine Cloud Platform Machine Learning In Financial Benefits and Ethical Enhanced Data Stewardship McAfee and Erik Andrew Considerations Services The Information Accountability Brynjolfsson Foundation Financial Board European Parliament Stability Outlook On AI in the AI in Payments: The Last Mile Intelligent Automation Humans Wanted in Efficiency Enterprise UBS RBC Finextra Pelican and NarrativeScience 156 The New Physics of Financial Services |

158 Acknowledgements

159 Acknowledgements | Contributors Contributors (1 of 6) and by through interviews perspectives ble The project team would also like to express its gratitude to the following subject matter experts who contributed their valua participating in workshop and roundtable discussions (in alphabetical order): 8 Securities Public Sector Pension Investment Board Mikaal Abdulla André Bourbonnais HSBC Raul Abreu SecureKey Andre Boysen Jeremy Achin DataRobot Motive Partners Jon Bradford Marchés Lise Estelle Brault Mukul Ahuja Financiers Autorité des Deloitte LLP, Canada James Breeze BBVA Elena Alfaro XL Catlin Dr. Larisa Angstenberger Macro Bressan AIM Satellogic University of Applied Sciences and Arts StreamLoan Thomas Ankenbrand Stephen Bulfer Lucerne Lloyds Banking Group Martin Arnold The Financial Times Claire Calmejane Anthony Bak Palantir Marcelo Camara Banco Bradesco Graeme Carmichael Deloitte LLP, UK Lee Baker Seldon Jo Ann Barefoot Barefoot Innovations Anna Celner Deloitte AG, Switzerland Rainer Baumann Swiss Re Management Soumak Chatterjee Deloitte LLP, Canada Santander InnoVentures Rohit Chauhan Mastercard Mariano Belinky RiskGenius Chris Cheatham Hallum Capital Group Craig Brad Berning - Ayan Bhattacharya Deloitte Consulting LLP, US DTCC Everitt - Marie Chinnici AMTD Group Alain Biem S&P Global Calvin Choi Michael C. Bodson Moses Choi Orange DTCC Greg Bonin XOR Data Exchange Diwakar Choubey MoneyLion Managers Nelson Chow Laurence Boone Hong Kong Monetary Authority Axa Investment Camp One Ventures Rob Claassen Victor Botlev Iris AI Andrew Connell HSBC Josh Bottomley HSBC 158 The New Physics of Financial Services |

160 Acknowledgements | Contributors Contributors (2 of 6) and by through interviews perspectives ble The project team would also like to express its gratitude to the following subject matter experts who contributed their valua participating in workshop and roundtable discussions (in alphabetical order): BT Adena Friedman Michael Cooper Nasdaq - Philippe Courtois Microsoft Jean Infosys Dennis Gada Jason Crabtree Jan Coos Geesink Fractal Industries Thomson Reuters Bisesti Gennaro Nasdaq Bill Dague Generali Insurance Avi Goldfarb Ray Dalio Bridgewater Associates University of Toronto, Rotman School of Management Kasisto Zor Gorelov Andrew Davidson Salesforce ComplyAdvantage Adam Gottlieb Charlie Delingpole True Accord Andrew Graham Clarity Money Adam Dell Borrowell Gero Gunkel Zurich Insurance Thomas J. DeLuca AMP Credit Technologies Ashish Gupta BT My Money Bank Alain Demarolle Reed Smith Paul R. Gupta Kees Dijkhuizen ABN AMRO Bank Daniel Drummer JP Morgan Chase RavenPack Peter Hafez Christian Faes Lendinvest Haenggi Marianne Zurich Insurance Autorité Financiers Marchés des Moad Fahmi RBS Kevin Hanley PingAn OneConnect Michael Fei Oli Harris JP Morgan Chase Fortia Sira Ferradans Luminoso Catherine Havasi Vickers Venture Partners Dr. Finian Tran Mark Hawkins Salesforce Dawn Fitzpatrick Soros Fund Management Jonathan Hayes Julius Baer Martin L. Flanagan Invesco Daniel Hegarty Habito De Montfort University Dr. Catherine Flick XL Group Greg Hendrick Barry Freeman Pintec Group Daiwa Securities Takashi Hibino 159 The New Physics of Financial Services |

161 Acknowledgements | Contributors Contributors (3 of 6) and by through interviews perspectives ble The project team would also like to express its gratitude to the following subject matter experts who contributed their valua participating in workshop and roundtable discussions (in alphabetical order): Yunfeng Financial Group Dianrong Gary Ho Ling Kong Dilip Rich Hochron Morgan Stanley LLP, US Touche Deloitte & Krishna Fabio Kuhn Jeff Holman Sentient Technologies Vortexa Ian Horobin Standard Chartered Bank Swift Shameek Kundu Angela Kwok Deloitte MCS Limited, UK Matthew Howard China Broadband Capital Swiss Finance & Technology Association Tradeshift Christian Lanng John Hucker IEX Group Stephan Hug Credit Suisse Laurence Latimer Patrick Hunger IDEO Colab Saxo Bank Ian Lee Sedicii Rob Leslie Timothy Hwang FiscalNote Shift Technologies The World Bank Jeremy Jawish Joaquim Levy Yunfeng Financial Group Aprivacy Cédric Jeannot Ting Li Deloitte MCS Limited, UK Gurpreet Johal Luca Lin Domeyard Mark Johnson Wayne Liu AMTD Group Descartes Labs BMO Financial Group Kevin Lynch Husayn Kassai Onfido Boris Khentov Betterment Simudyne Justin Lyon AIG Sunil Madhu Socure Reza Khorshidi of London Lloyd’s Haq Shirine Khoury - Neeraj Makin Emirates NBD Westpac Tameryn Mallet Fiscal Note Martin Kilmer UAE Exchange Promoth Manghat Silicon Valley Bank Dan Kimerling Christer Kjos Jason Mars Clinc Canica International Deep Science Ventures Inma Martinez Michael Kollo AXA Rosenburg 160 The New Physics of Financial Services |

162 Acknowledgements | Contributors Contributors (4 of 6) and by through interviews perspectives ble The project team would also like to express its gratitude to the following subject matter experts who contributed their valua participating in workshop and roundtable discussions (in alphabetical order): Lloyds Banking Group True Wealth Vim Maru Felix Niederer Zendrive Jonathan Matus Third Point Capital Matt Ober Máxima of the Netherlands Dutch Royal Household RenaissanceRe Holdings O’Donnell Kevin DTCC Robert Palatnick Ofer Mendelevitch LendUp Autorité Marchés Financiers Francois Mercier Fountainhead Partners Harry Pang des Meyer Lending Club Suade Labs Diana Paredes Armen Mark Patterson Greg Michaelson DataRobot Deloitte MCS Limited, UK Guggenheim Partners Ana Pinczuk Jerry Miller Hewlett Packard Enterprise Adecco Hans Ploos van Amstel Ross Milward Quantifeed Mauricio Minas Banco Bradesco Jayne Plunkett Swiss Reinsurance Company Rebecca Minguela Clarity Christina Qi Domeyard Hans Morris Andy Rachleff Wealthfront Nyca Partners ING Netherlands Vinoth Raman Craig Morrow Atom Bank James Ramirez Sophia Mouhoub Essentia Analytics Fortia Old Mutual Emerging Markets Peter Moyo Aki Ranin Bambu Brody Mulderig Wells Fargo Arun Rao Starbutter Science - Sunil Rawat Berkman Center at Harvard University Patrick Murk Omni Prakash Nanduri Paxata Anne Richards M&G Investments (Prudential Financial) Krishna Nathan Ellen Richey Visa S&P Global Hans - Prudential Financial Rieder UBS Michael Natusch Juergen Schroder ADVEQ Nils Rode Nguyen Thomson Square Capital 161 The New Physics of Financial Services |

163 Acknowledgements | Contributors Contributors (5 of 6) and by through interviews perspectives ble The project team would also like to express its gratitude to the following subject matter experts who contributed their valua participating in workshop and roundtable discussions (in alphabetical order): Generali Shift Technologies Philippe Roger Donnet Stephanie Steele Liberty Mutual Anne Rush Betterment Jon Stein Kirsty Rutter Barclays UK ForwardLane Nathan Stevenson Homer Strong Koby Sadian Cylance Viking Global Investors Shalini Sujanani J. Safra Group Jacob Safra ING Bank Cambridge University Point72 Asset Management Michael Sullivan Mark Salmon Insurance Ohad Samet True Accord Zurich Tom Swaan Satyen Sangani Alation Malcolm Sweeting Clifford Chance Gert Sylvest Gautham Sastri iSentium Tradeshift Stefan Teis Deutsche Börse H2O.ai Sri Satish Ambati Andrew Schlossberg Invesco Perpetual Andy Tong MPFA Cameron Schuler Alberta Machine Intelligence Institute Carlos Torres Vila Banco Bilbao Vizcaya Argentaria Jennifer Zhu Scott CitiVentures Radian Partners Luis Valdich Rob Sears BBVA Aire Aneesh Varma Michael Shaver Deloitte LLP, Canada Compagnie Financière Tradition Dominique Velter Brian Walnut Algorithms Deloitte Consulting LLP, US Shniderman Guillaume Vidal Sanjaya Shrestha Ajay Vij Infosys Barclaycard Travis Skelly CitiVentures Anna Wallace Financial Conduct Authority Credit Suisse Lara J. Warner Decipher Finance Sean Slotterback Mov37 Michael Weinberg David Spreng Crescendo Ventures Matt Wettlaufer Scotiabank John Stackhouse RBC 162 The New Physics of Financial Services |

164 Acknowledgements | Contributors Contributors (6 of 6) The project team would also like to express its gratitude to the following subject matter experts who contributed their valua ble perspectives through interviews and by participating in workshop and roundtable discussions (in alphabetical order): Andrew Patrick White FundApps Andrew White Wealthfront WilmotML Aric Whitewood eTrading Software Grant Wilson Greg Wolfond SecureKey Marathon Ventures Ray Yang Tohmatsu Certified Public Accountants LLP, China Deloitte Jennifer Yi Qin Touche HSBC Janet Yuen Snejina Zacharia Insurify Ryan Zagone Ripple Brian Zboril CME Eunice Zehnder IPM Bryan Zhang Cambridge Centre for Alternative Finance Credit Ease Nina Zhou Martin Zielke Commerzbank Ori Zohar Sindeo 163 The New Physics of Financial Services |

165 Contact details

166 Contact details lease ontact: For additional questions, p c SUPPORT FROM WORLD ECONOMIC FORUM PROJECT TEAM PROFESSIONAL SERVICES DELOITTE McWaters R. Jesse Rob Galaski Project Lead, Financial Services Leader, Banking & Capital Markets Global Deloitte Consulting World Economic Forum [email protected] [email protected] Matthew Blake Courtney Kidd Chubb Senior Manager Initiatives Head of Financial & Monetary System World Economic Forum Deloitte, Canada; [email protected] m [email protected] Denizhan Uykur Senior Consultant Deloitte, Canada; [email protected] 165 The New Physics of Financial Services |


Related documents