Deutsche Bank
Artificial Intelligence from the perspective of a Retail Bank Jürgen von der Lehr, Managing Director Head of Digital Innovation & Think Tank Deutsche Bank AG, Private, Wealth & Commercial Clients 30 MAY, 2017
Contents
A
Artificial Intelligence is now ready to be used in broad application and offers solutions to retail banking’s current challenges
B
DB’s journey in AI: Verbal Update
C
Lessons learned and our outlook in AI
Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
1
Artificial Intelligence (AI) is now ready to be used in broad application Development of Artificial Intelligence
Why is AI important now?
1960-1980
• Scale of computational resources – More computer power – Cheaper computer power
now
Great promise AI’s rebirth to revolutionize, but no delivery 1956 Theoretical foundation
• Amount of big data available to our machines – Learningsystems get betterthe more datathey use • Adoption of alternative reasoning models – Systems do not have to reason like people
• Shift away from the idea, that AI has to solve everything – Systems like Siri and Cortana work better within limited domains of action – Narrow financial search engines such as Kensho answer very specifically to very specific questions
Artificial Intelligence: Machines that are able to mimic “cognitive” functions that humans associate with other human minds, such as “learning” and “problem solving”. Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
2
AI offers solutions to retail banking’s current challenges
Today’s retail banking industry challenges …
… are met by the value proposition and main benefits of Artificial Intelligence
Change in client behavior (e.g. expectation of instant feedback)
Transformation of user / customer journey without human interface
Large amount of unstructured data remains unused
Empowered relationship managers with usage of unstructured data
Cost pressure on operations (e.g. number of legacy systems)
Increased efficiency in operations, via selfmanaging systems
Regulatory pressure increases complexity in processes
Automated processes to save time and effort on regulatory demand
Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
3
AI will disrupt the banking industry much like other industries have experienced disruption
Disruption in other industries…
…and their similarities to AI Artificial Intelligence (banking):
3D printers (manufacturing): Become independent of logistics Enable instant feedback
Autonomous driving (transportation): Become independent of human error Enable strictly standardized results
Service without human interface Instant feedback
Standardized results without human error Smooth experience
Source: Deutsche Bank; Oliver Wyman
Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
4
In banking, AI can provide an intelligent financial assistant for clients 1. INTELLIGENT AI EXPERIENCES From complex questions to personalized recommendations Key examples: • “How can I achieve my goal of …?” • “Show me how to improve my spending” • Key triggers (e.g. baby) > Personalized recommendations • I am not happy with this service… should I change bank?
AN INTELLIGENT ASSISTANT THAT
HELPS ORGANIZE AND IMPROVE YOUR FINANCES
AND CONNECTS THEM WITH LIFE BEYOND BANKING
IN A NATURAL AND SEAMLESS EXPERIENCE
Deutsche Bank Private, Wealth & Commercial Clients
2. KEY BANKING FEATURES Necessary banking features and offerings Key examples: • Account dashboard • Savings • Transfers & Payments
3. BEYOND BANKING AREAS Connection with other aspects of life Key examples: • Travel (e.g. plan perfect itinerary) • Retail (e.g. get tailored deals) • Housing (e.g. improve utility spending)
4. MULTI-MODAL / MULTI-CHANNEL EXPERIENCE Across modes (visual, voice, text) & channels Key examples: • Apps • Amazon Alexa • Desktop (now) / wearable (future) • Chat
Jürgen von der Lehr May 30, 2017
5
Banks are currently significantly investing in different AI solutions to be the front-runner in that space Non-exhaustive list
Most banks are in early stages of adopting AI technologies (peer usage of AI solutions, examples) • Search employee communication against insider trading • Model behavioral patterns of wealthy clients to WM products • Self-service customer services • Prepare more efficiently for client meetings • Support their Trading Record Keeping Compliance solution • AI customer assistant to help staff answer customer queries • Near real-time customer insights to prepare client meetings • Web assistant to handle >350 customer questions & answers • 360° view of client data • Speech & text recognition for call center automation • Improve client investment suggestions
Current market view • Leading banks understand AI as competitive advantage and invest significant amounts • Most solutions currently still in pilot stage – reasons being high costs and lack of internal AI expertise • Speed and complexity of implementation often driven by: – Complexity of use cases – Parallel or sequential implementation of pilots – Usage of external providers – Dependencies on IT-legacy systems
• Interpret and maximize use of customer data • Integrate contextual external sources information
Access to AI expertise is essential for establishing strategic advantages. Sources: IBM; Company websites; Wall Street Journal; Süddeutsche Zeitung; Oliver Wyman
Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
6
Observation: Use cases in AI typically cluster around three interaction counterparties … Use case counterparty
Direct client experience
Client advisor productivity
Role of AI system Enable direct interaction between client and AI system, e.g. for small advisory tasks
Free client advisor resources and enhance advisory quality, e.g. in client meetings preparation
Potential use cases (illustrative, not exhaustive)
• Self-directed user advice • Q&A guided KYC
• Intelligent advisor tool to free client advisor from preparation work • Automate complaint processing to relieve client advisor • Guide internal advisor services hotline
Back-/ middle-office efficiency and enhanced decisioning
Automate manual processes with multiple stakeholders, e.g. in client onboarding
• Financial crime transaction monitoring with machine learning feedback loops • Trader surveillance to improve conduct-related controls with machine learning file analysis • Automated handling of mis-selling claims
Use cases critically dependent on acceptance within organization – unguided, direct, client interaction usually requires stronger internal acceptance Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
7
Usage of AI goes beyond classic retail banking – especially capital markets and asset management have been active Emerging applications
Example firms
Description
Listed asset trading
• AI-based hedge funds using varied underlying AI technologies to trade listed liquid assets and derivatives including commodities • New firms and incumbents developing in-house technology
Commodities trading
• AI to decipher satellite imagery to make forecasts (e.g. oil inventory, crop yields) for potential use in trading • Investors include Bloomberg, Sequoia Capital, Google
Risk modelling
• Analytics based on topological data analysis • Investors include KPCB, Khosla Ventures, Citi Ventures • Citigroup and Credit Suisse among banks using CCAR stress test app
Research, analytics and report writing
• Natural language processing and generation to aid sales and trading • Kensho backed by Goldman Sachs, Google, CNBC with JPM, BAML as clients • Narrative Science used by banks (e.g. CS) and asset managers
Client engagement and assistance
• IBM partnering with banks to deploy Watson to assist client relationship managers (e.g. ANZ, DBS) • Allows tailored advice and offerings for private bank clients
Employee oversight
• NLP-based analysis of communications to raise early warning flags • Investors include Goldman Sachs, Credit Suisse NEXT Investors • Alliance with NASDAQ to offer e-communication monitoring
Cyber security
• AI-enabled cyber threat prevention and detection in real-time
Source: Oliver Wyman
Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
8
Contents
A
Artificial Intelligence is now ready to be used in broad application and offers solutions to retail banking’s current challenges
B
DB’s journey in AI: Verbal Update
C
Lessons learned and our outlook in AI
Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
9
Contents
A
Artificial Intelligence is now ready to be used in broad application and offers solutions to retail banking’s current challenges
B
DB’s journey in AI: Verbal Update
C
Lessons learned and our outlook in AI
Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
10
Lessons learned: Importance of setting up a functional organizational governance and of embracing strong planning Lessons learned in AI from leading financial services institutions
New technology needs internal buy-in and acceptance ─ At top-management level: To ensure high priority on technology roadmap for implementation ─ At business function / client advisor level: To ensure actual usage of new technology ─ Business is more likely to accept use cases that relieve work from internal resources or that guide decisions than to accept use cases that take over parts of client advisory and that put jobs at risk Initial Proof of Concept (PoC) phase makes up leeway at later implementation stages ─ Get time to understand benefits of new technology to generate acceptance ─ Get time to figure out smooth internal governance and responsibilities of implementation ─ AI is a scalable technology that allows for module-based add-ons on demand
AI systems are only as strong as they have been trained for ─ Time-consuming (multiple months to years) ─ Training needs real and diverse data that is suitable, i.e. understandable ─ Even after training, AI will not necessarily give “black-or-white-answers” Centralizing internal AI competence allows for oversight and orchestration of AI activity ─ Avoid multiple divisions to work on the same ideas by giving clear guidance
Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
11
Artificial Intelligence is a multi-year journey with the potential to disrupt the way we think banking today The disruptive potential of AI …
… will transform retail banks
The rise of powerful AI will be either the best or the worst thing ever
Transformation of user / customer journey without human interface
to happen to humanity.
Empowered relationship managers with usage of unstructured data
We do not know which.
Increased efficiency in operations, via self-managing systems
– Stephen Hawking
Automated processes to save time and effort on regulatory demand Deutsche Bank Private, Wealth & Commercial Clients
Jürgen von der Lehr May 30, 2017
12