What Financial Services Firms are Doing with AI Today Will Separate Leaders from Laggards

As an industry analyst and thought leader, I’ve attended many great discussions on the role of data, analytics, and artificial intelligence in industries and organizations. Today’s virtual event with The Economist on how AI is transforming financial services ranks among the all-time best I’ve participated in.

Here are the top takeaways from the panelists, what separates the leaders from the laggards, and my takes and recommendations.

Financial Services is Most Advanced in AI

Katya Kocourek, Managing editor of The Economist Group kicked off the event with a pointed question: how useful is AI for the financial services industry? Particularly amid the global economic crisis, the Wall Street Journal suggested a “chilling” toward AI and scrapping of some projects. Only last year, Gartner predicted half of AI projects remained “alchemy run by wizards.”

Manuela Veloso, Head of AI research, J.P. Morgan, took a more reasoned stance, noting that AI is still a relatively young field, but that it is most mature in financial services because of the vast amount of data available in this industry. Sally Eaves, Official member, Forbes Technology Council, cited research that the APAC region has a 20 percentage point differential in extracting value from AI compared to North America and EMEA. As COVID continues to reshape our working world, this is only accelerating. 

My take: In my work in leading the data and analytics maturity models at Gartner, financial services was one of the most mature. This was echoed by recent research we did with Harvard Business Review. Without data, one cannot do AI, so the industry’s advanced maturity relative to other industries isn’t surprising. Singapore and Malaysia have shown leadership in AI, early with ethics guidelines and data literacy investments in society and education; for Singapore, the same could be said of their investments in broadband for every citizen just as the Internet was taking off in the USA. A large part of Singapore’s economy depends on financial services. China, meanwhile, is in a race to beat the US in AI innovation. 

No Business Value? Don’t Bother!

Part of the struggle with AI is in finding the right use cases for AI, then proving there’s value in applying AI to it. 

Terry Hickey, Chief Analytics Officer, CIBC, recommended the top three use cases for financial services:

  • Reduce risks. Protecting the bank is paramount and above all else. This includes fraud detection and credit risk. 

  • Improve revenue. Expanding the bank's share of wallet through cross and upselling is a critical path to growing revenue.

  • Decrease costs. By automating processes, banks can lower their operational costs and drive the bottom line.

Financial services organizations need to then look at the ROI across these three use cases. Hickey added that unless one can show the benefits, the money to fund AI projects will dry up.

My take: Organizations must indeed align to business outcomes for any AI and data effort. I include this as a key pillar in the ThoughtSpot data and analytics maturity model. It’s one reason our business value consultants engage with customers earlier in the software sales cycle. However, the ability to quantify ROI has been an ongoing debate in the industry. I taught a class on this topic at TDWI for many years, devoted a chapter in one of my books, and further studied at Gartner. The only part that is easy about calculating ROI for such a project is the cost or denominator; ROI is a precise number with imprecise inputs. The numerator or benefits is often up for interpretation. Did revenue increase because AI-driven insights improved customer service and cross-selling? Or was it because the bank provided better training of financial advisors? 

My recommendation? Capture and cite anecdotal business benefits of an AI initiative. Use a back-of-the-envelope calculation to be sure AI is driving value and not falling into wasteful research projects.

Trust is a Prerequisite for AI.

Eaves cited research from Edelman recognizing that consumer trust in business is at a 17-year low. As AI becomes more and more prominent in financial services, there’s a real possibility this could get worse. Distrust increases when people do not understand AI or the inputs that drive a model.

To combat this, organizations should take two steps: make AI more transparent and explainable, and focus on diversity. The more consumers and users can understand your technology, the more they’ll trust it. Diversity is also critical. Organizations need to look at diversity of experience, advised Eaves. Diverse teams are more innovative and creative, and they create a more holistic impact. This kind of social impact is a way to foster trust. J.P. Morgan’s Veloso added that there is a responsibility to use data to train AI that’s representative of our diverse world and captures it’s complexity. Dave Oliver, Head of nerve centre, RBS, noted that the way a science student might evaluate building an atom bomb will be very different from how a humanities student would approach this. Both approaches are important for a holistic understanding of nuclear weapons. The same multifaceted view is needed for AI. 

My take: Never having met these speaker’s before, what can I say? Great minds think alike! See my recent discussions on diversity and data for good.  

Covid is a Forcing Function for AI

We are only several months into the severe impacts on the economy from the global pandemic, and yet, the experts at the event agree there will be profounding, lasting changes to the finance world as a result. For example, the need for external and more timely data has never been greater. Manuela used the term “nowcasting” as opposed to forecasting. Hickey predicted that conversational AI and NLG will be of greater importance as we move to non face-to-face interactions. The best of AI continues to combine human intelligence with machine intelligence. Hickey used a perfect analogy here: autopilot might help fly a plane, but would have been unable to land a plane safely on the Hudson River like Pilot Sully did. The importance of AI in financial services, like transportation, is equality important. As Manuela saide, “it’s not about losing a chess game; it is the lives of people - deciding on a loan or a credit. This is so much more than an aid for a recommendation. It’s a serious industry.”

My take: As I said in the webinar in the early weeks of this pandemic, speed to insight has never been more important. Like it or not, ready or not, COVID is a forcing function. Identifying credit risk at a granular level, in a scalable way is paramount; trying to do this in spreadsheets for large banks is impossible, with risk of errors in CECL reporting. Any financial services company who either sloppily or inadvertently threatens a newly unemployed customer, will never earn their loyalty back. That is at best. At worst, I fear the social harm we’ll inflict if we do not bring the human heart into these financial decisions. To truly service these customers, we need better leading indicators that external data sets provide. Three options include data sources from Eagle Alpha, FactSet, and Unacast for human mobility.

If you’re a financial services leader ready to accelerate your journey with AI, I encourage you to check out the full report from The Economist, and their collection of resources for companies ready to unlock the value in all their financial analytics data. Personally, I hope that’s everyone. Otherwise, we are just adding to a digital wasteland.