Dr. Christoff Nieuwoudt delivered a keynote on AI in Financial Services at Digital Finance Africa 2023 on the 2nd of August 2023 at Gallagher Convention Centre, Johannesburg, Midrand.
1. AI in Financial Services
Digital Finance Africa
Christoph Nieuwoudt
2 August 2023
2. 2
CONTENT
• The evolution of Artificial Intelligence
• Application of AI across industries
• How we are structured in the group
• The Role of Generative AI
• Summary and what is next
3. 0,6 1,0 1,5 2,3 4
6
0,1 0,5 3,1
9,3
21
34
0
5
10
15
20
25
30
35
40
1975 1985 1995 2005 2018 2023
Tangible Intangible
Massive growth in company valuations driven by technology, data, IP and other
intangibles in 3rd and now into 4th industrial revolution
Source: Statistica; Gurufocus.com; Bloomberg; Team estimate
ESTIMATE
IBM
Exxon Mobil
Procter & Gamble
GE
3M
IBM
Exxon Mobil
GE
Schlumberger
Chevron
GE
Exxon Mobil
Coca Cola
Altria
Walmart
Exxon Mobil
GE
Microsoft
Citigroup
Walmart
Apple
Alphabet
Microsoft
Amazon
Facebook
Apple
Microsoft
Alphabet
Amazon
Nvidia
Top 5 by
market cap
Tangible vs Intangible Assets for S&P 500 Companies, 1975-2023
USD trillion
4. 4
The shift to Data & Analytics is structural and is expected to continue for the
foreseeable future
*This belief was/is widely held including by Stephen Hawking, Yuval Harari, Bill Gates, Elon Musk, Larry Page, Sergey Brin. Ray Kurzweil puts it at 2029, but most commentators at 2040 or
after. “Companies have to race to build AI or they will be made uncompetitive” Elon Musk ; “AI-driven companies will take $1.2 trillion from competitors by 2020” Forrester.
Source: Ray Kurzweil, DFJ; International Data Corporation Data Age 2025 Study; Team analysis
1
10
100
1,000
2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025
Global Data Growth
(ZettaByte = 10^21 Bytes)
The combination of a sustained exponential increase in computing power, which has lasted for 120 years (60 years since Gordon Moore’s
article) and close to exponential increase in data generation along with algorithmic improvement drives an ongoing shift in competitive
advantage towards data and analytics-based firms as the world nears the point of machine intelligence exceeding human intelligence*
Algorithmic evolution is the third leg of ongoing improvement driving the influence of machine learning – e.g. “deep learning” improved neural-network performance to the point where
computer vision and speech nears and for many applications exceeds that of humans. Reinforcement learning algorithms defeats humans in all major competitive games. Generative AI
is the latest development, unlocking knowledge itself by integrating a large portion of human writing, code development as well as images, video and audio.
5. Generative AI is the latest development in AI, building on developments in Deep
Learning and Reinforcement Learning and applying it to text, images, video and other
data
Source: Microsoft
6. AI value is great not just in financial services, but can greatly benefit our
customers & ecosystems across all sectors
Source: McKinsey & Company articles (www.mckinsey.com) - see references in appendix; CDAO views
Areas of direct interest
7. Our Data & Analytics strategy focusses on value
across time horizons
• Build foundational group data products
• Customer data asset (integrated by FirstID)
• Transactional data assets (2.5bn pmts, R50tn)
• Interaction data (~2bn monthly interactions)
• Credit, Invest, Insure data assets
• Ecosystem data assets (Home, Car, External)
• Design & deploy models for insight driven decision
making and efficiencies
• Customer targeting
• Cross-sell optimization across Transact, Lend,
Invest, Insure
• Credit, Crime (Fin Crime, Fraud), Collections &
Capital (including insurance underwriting)
• Cost reduction & work-item automation
• Leverage and enhance data products to offer
bespoke, contextual customer value
propositions
• Retail – Integrated advice and product/
solution offerings leveraging NLP driven
interactions, robo-advice & “green”
tools
• Commercial & Corporate – Client
advisory services based on deep client
profiling
• Externalise key capabilities to clients,
partners and stakeholders
• Build full platform ecosystems beyond
traditional banking, but integrated &
leveraging banking solutions to bring Retail,
Commercial, Corporate and Institutional
clients together
• Marketplace bringing buyers and
sellers of goods together with deep
contextual information to optimize
offers and marketing
• Build ecosystem around “home”-
residential & commercial
• Build ecosystem around “car”- OEMs,
groups, dealers, buyers & sellers
including B2C, B2B and C2C
marketplace
• Tap into & develop ecosystems across
adjacent markets to offer seamless,
intelligent solutions
Build the data foundations to do better what we
already do …
Leverage data foundations to materially enhance
our VP including client advisory services …
Build a platform business linking buyers and
sellers via marketplaces and ecosystems
Finance-only offerings are becoming
commoditised in the medium-term, which the
ability to offer deep advice and externalize our
capabilities can counteract
Platform ecosystems unlock value for all
participants, making it the natural long-term
winning business model across most industries
Source: Internal interviews (CEOs across Brands, Segments, BUs; Brand/Segment and Pillar CDAOs; Risk/Legal/Information Governance), Literature review, Team analysis
Aim: Improved financial
services
Aim: Data as a new
business
Aim: Marketplace and
Ecosystem business
8. CDAOs play a strategic role in every segment, pillar, domain and function
across the group and are critical for AI delivery
Retail
Commercial
Corporate &
Institutional
Transact, Lifestyle
& Connect
Lend Invest Insure FirstID Interactions
PILLARS DATA DOMAINS
Risk Finance
FUNCTIONAL AREAS
Human
Capital
Broader
Africa &
Aldermore
Source: CDAOs
Data and Analytics Platform Information Governance &
Architecture
Programme
Management
Solution
Architecture
Treasury
Decisioning
Platform
9. GAI tools are disruptive as they can easily be deployed, integrate large language,
code, conversation and image/sound capabilities enabling key use cases
*Useful, but not necessarily accurate, up-to-date etc.
Source: Microsoft
GAI can be easy to use and useful* “off the shelf”
GAI covers text, code, conversations, images and more
GAI supports a number of use cases
in financial services
10. Private instance(s) are required to address privacy/PI and confidentiality/IP
concerns and use of embeddings can address accuracy concerns (including
bias, ethics etc.)
Source: Microsoft; CDAOs
To provide ‘accurate’ answers based on bank information, approach is to embed
(index, split & vectorize) reference documents and make sure answers supplied are
based (only) on own information
A private instance using pre-trained models ensures
privacy and confidentiality with ability to fine tune model
usage and implement own controls
Illustrative example, but the same logic & architecture would apply to use of AWS, Google Cloud or
other GAI tools, including open source LLMs and related tools
11. McKinsey projects USD3-4tn in additional benefit from GAI (already USD11-18tn
p.a. across AI use cases), which our use cases largely align with
Our initial use cases for GIA align nicely with the top
potential application areas identified
• Search & chat improves customer operations by
automating key work items in each segment on
Navi (our chat engine, including voice). We
already do this, but with GAI we can do it so much
better.
• Personalised communications and advice
combine marketing & sales by addressing how
products and solutions are positioned to each
customer including wording and imaging.
• Code assist – software engineering comprise the
top use cases and we will pursue across different
coding applications
• Report automation & other efficiencies seeks to
leverage this across a range of use cases –
improving on what we already do in Fin Crime
reporting.
• Staff productivity tools. AI and GAI is being
incorporated by a number of vendors to make GAI
available in staff working environments
12. 12
SUMMARY AND WHAT NEXT
• Further enhance data product development, model governance and ethics for
AI/GAI
• Further development of platform capabilities for ML/AI/GAI Ops with acceleration
of model deployment and ongoing cloud migration
• Enhance existing AI use cases (Fincrime, Fraud, Biometrics, OCR, etc.) while
prioritising new GAI enabled use cases
• Upskills CDAO/wider D&A community & AI Literacy for all staff across group
Notes de l'éditeur
Grossed up to SA Industry level using Issuing Market Share