When AI meets banking
“Can machines think?” — a simple yet powerful question that computer pioneer, Alan Turing, first brought up during the dawn of computing in the ’50s. It sure made the naysayers frown at the unthinkable. Yet today, artificial intelligence (AI) is everywhere — from an AI assistant robot in space, to it being a handy companion in all our smartphones. Its proliferation has only made our work and personal lives easier.
We had a chat with Colin Priest, Vice President of AI Strategy at DataRobot, who discussed how AI has helped us harness our collective intelligence. He also shared more about working with DBS through the DBS Startup Xchange programme to improve processes in Hong Kong with AI.
Colin, can you tell us all about your collaboration to use DataRobot’s AI with DBS?
The financial services industry has become increasingly complex and competitive as customers grow accustomed to instant gratification and on-demand services. We have been working with DBS to improve operations and processes with the integration of AI solutions. DataRobot has been helping DBS make the most of its talented banking team, by filtering out information overload, freeing up staff to focus on valuable issues that require human attention.
For example, DataRobot can predict and identify delinquent customers who are at higher risk of not meeting their financial obligations. This insight will allow the bank to take preemptive action. As more customers turn to digital channels for their banking needs, we are also building a solution to automate loan approvals for Hong Kong SMEs, which will enhance the overall customer experience while also improving risk quality.
From Risk Management Group (RMG) to Treasury and Markets, Audit and Finance — we are engaged with multiple business units within DBS to transform, simplify and improve operations and processes. Some of the AI-solutions that are currently in development include a tool that can predict the yield of a newly-issued bond and a complaint prediction tool that helps to proactively identify and resolve issues, as well as an end-of-day balance and cash-flow prediction tool that is still at the initial stage of ideation.
What is unique about DataRobot’s solutions as compared with other AI software applications in the market?
2020 is the year that AI has become a mainstream enterprise technology that can reliably deliver bottom-line results. However, without industrialisation and democratisation of AI, many organisations will fail to achieve AI success.
DataRobot is unique in its ability to automate much of the process of building and operationalising AI. We also apply safety guardrails to prevent errors and proactively monitor AI health.
Further to that, we provide user-friendly and easy-to-understand explanations to our AI solutions, which makes it highly accessible to businesses. We want to empower organisations to build their own AI models rather than rely on black-box proprietary software. By building your own AI models, you build trusted AI that is intuitive and works as planned.
What do you think about businesses’ adoption of AI? How can it help to drive results?
There is a lot of hype surrounding AI which makes it seem far-fetched like science fiction. However, the value of AI is real and its impact has been demonstrated across industries.
Research shows that AI is unlocking billions in value each year and that AI-powered businesses have a competitive edge over others. Many are also saying that late adopters of AI may never recover from lost opportunities. Yet, businesses are still holding back because of the lack of understanding and simply don’t know where to begin.
We do not believe that technology should drive business. Rather, technology should serve the needs of a business. AI is one of the most versatile technologies humans have ever developed. Even though there have been hundreds of uses for AI, I still manage to discover new ideas each day.
For example, I have been surprised by the new use cases and ideas that people came up with during the AI workshops that I regularly conduct for business managers. Often, it is about improving the customer experience, preventing errors, or improving the efficiency of a process. But I have also seen simple yet successful AI projects that prevent forest fires, repair power lines before they fail, and even predict which patients will miss their medical appointments!
What were some stumbling blocks faced during the project with DBS, and how did you overcome them?
Two hurdles confronted DBS and DataRobot during our time at Startup Xchange in Hong Kong. The first was honing in on the AI opportunities as we had hundreds to choose from. The challenge then was agreeing on the proposition to undertake.
The second challenge was identifying the right building blocks for a successful implementation. While some of the project decisions were technical, others were focused on customers, organisational change, regulations, and even practical business workflows.
We eventually overcame these challenges by putting together a well-rounded team of experienced data scientists, AI success managers and business subject matter experts. Together, the team ensured that the AIs were technically feasible and aligned with business objectives. Most importantly, they were adopted by the businesses to accelerate the path to value.
What were some interesting insights or discoveries gathered from this project?
The DBS Innovation and Startup Xchange team in Hong Kong have been a great support throughout the process. Through this collaboration, we have gotten the various business units excited about AI and making its use pervasive throughout the organisation. In fact, the Startup Xchange team were so enthusiastic in driving AI success that there were instances where they were mistaken for DataRobot employees!
There is one notably fascinating use case that was discovered through the collaboration between our team and Startup Xchange. The Treasury and Markets division faced a problem as debt financing was starting to dry up and companies were starting to look towards capital markets for financing. This means that predicting the yield of a new bond becomes increasingly important. To solve this problem, we were able to assist the team in the development of an AI model that predicts the yield or “G-Spread” of newly issued corporate bonds.
What role will be emerging technologies like AI play in shaping the future of banking and finance?
Banks have been at the forefront of using technology to automate and increase the speed and scale of their operations. I remember the days before computers were invented — everyone had bank passbooks and each transaction was recorded manually by the bank teller. Now, I can bank from my phone and obtain a realtime view of all of my transactions and bank balances.
I think that what we are going to see with AI is more intelligent automation. This means smarter automation that can do more complex tasks and personalisation at scale for customers. A good example of this is our collaboration with DBS to automatically underwrite loans for SMEs. This level of automation wasn’t possible in the past where rule-based decision was not able to accurately capture the complexity and nuance of different business sectors.
AI will also improve banks’ ability to be agile. While the coronavirus has drastically upended the world we live in, AI can discover and anticipate future trends. Rule-based decision making cannot adapt to dynamic and rapid changes, but AI can notify us when the old ways are changing. Overall, AI can offer customers a frictionless experience, enhancing speed and accuracy in providing the right products, preventing fraud errors and better anticipate unexpected events.
