star-bg.jpg

CognitiveScale Blog

Ethical and Trusted AI in Financial Services: CogX 2020 Panel Recap

By admin Jul 10, 2020 11:40:51 AM

This year’s CogX—one of our favorite events of the year—was, even while virtual, one of the largest, most inclusive and forward-thinking online gathering of leaders, CEOs, entrepreneurs, scientists, artists, and activists in the world all focused on the huge theme of 2020: “How do we get the next 10 years right?” The goal was to focus on moving conversation forward with concrete actions, inspire current and future generations of leaders, and help to see the climate emergency and the current pandemic as two enormous opportunities for innovative solutions. 

 

In the panel discussion hosted by our own Manoj Saxena, Chairman of CognitiveScale, HSBC’s Gavin Munroe—Global CIO of Retail Banking and Wealth Management— and Ranil Boteju—Global Head of Data and Analytics, Retail Banking and Wealth Management—join him to discuss trusted AI in the financial services and banking space. The three main areas of focus:

  1. What does ethical decisioning mean to retail banking?
  2. What are opportunities for enabling this for the bank? Every crisis is an opportunity.
  3. Insight on how HSBC is taking on these challenges. 

To kick off the panel, Saxena explains his perspective on what it means to be trusted and ethical when it comes to automated decisioning systems. “Trust,” he says, “is the foundation of the digital economy.” While automated insights are being woven into every app, cloud workload, and data center, many are what we at CognitiveScale see as black boxes. Being digital requires embedding Trust As a Service into all digital experiences to drive transparent and ethical engagements. This, to us, means businesses focusing on management of six key dimensions of AI: explainability, bias and fairness, robustness, compliance, data risks, and effectiveness. The need for trusted AI, especially in a post COVID-19 world, is only going to increase.

So, what are some of the key COVID-19 customer challenges and opportunities? Munroe explains the basic difficulties customers face not being able to get to a bank branch, transfer funds, or make loan payments. The digital footprint has a lot of diversity, and some countries are mandated to close branches, others must offer various loan programs, and so on. HSBC has the responsibility to expose all of these things to their customers. “Obviously,” Boteju explains, “there’s a severe disruption right now.” It’s a time to try and use data and analytic tools to route people in the right direction. Some rules have to be tweaked in order to not end up with a bunch of false positives in terms of fraudulent behavior. It’s a time of transition and seeing what works.

There’s this misconception that building trust with the user means sacrificing your fintech IP, and all three agree that is just not the case. As Saxena explains, you don’t need to provide internals but rather evidence with explanations. Instead of saying “your loan got denied,” every solution should have the ability for algorithms to provide circumstances in which it would not have been denied—a certain income or another quantifiable factor. Boteju added that his advice on this was to have “an Ethical True North.” There is a need for tighter governance, and of course, there is a fine line of explainability. 

While there was talk of how to begin tackling all of these huge transitions—an education component, activation, scaling, etc—Boteju simply expresses that there is a lot of work to do. It still takes manual muscle power to get this all through and the use of AI will only accelerate it. The next decade will be a year of transition from exploring the “.Com architecture” to the “.AI architecture,” as Saxena likes to call it, and it will essentially include reimagining data, structure, and experience. One solution will never fit all, though making AI workloads portable across regions and clouds will be a huge step.

Nearing the end of the panel, both Boteju and Munroe express the desire for people to trust HSBC with their data in the same way that they trust them with their money. Building on trust, scaling on what we all already have, and leveraging upon a new norm. This is what will begin the great achievements of the next 10 years.