Every AI project should have an ethics committee that reports directly to the CEO and ensures that algorithms pass the 'common sense test', writes Brian Caplen.

The best banks of the future will be those that most effectively use big data and analytics by way of AI and machine learning. But there are many pitfalls on the road to success and banks that ignore the dangers may find themselves generating the wrong kind of headlines.  

Any kind of customer profiling that determines access to credit or financial products, and that so much as appears to have a gender bias, or indeed any other bias, is an obvious example of an accident waiting to happen. Any opaque system generating results that appear to be correct – but which the AI engineers have trouble explaining how they came about – should also be a cause of concern

Then there are worries about the sources of data and its security. More data should produce better outcomes, but if the data comes from social media or other external sources, huge care must be taken to verify its accuracy. There is also the question of keeping this data safe as leaks will start to become as reputationally bad for banks as having funds stolen.

The solution to all these challenges is not better machines but better humans. Every AI project should have an ethics committee that reports directly to the CEO as well as a process in which smart people interrogate the AI experts with ‘dumb’ questions to be sure that things are on a common sense track.

AI is a powerful tool with huge scope so there is an obvious risk of bank CEOs being blinded by science and going along with algorithms that offer results, even if the explanation as to how they achieve them is not entirely clear. For sure some banks will get this all wrong, enabling others to learn by their mistakes. CEOs need to plan carefully if they want to avoid being a lesson for their competitors.

Brian Caplen is the editor of The Banker. Follow him on Twitter @BrianCaplen

Register to receive my blog and in-depth coverage from the banking industry through the weekly e-newsletter.

What impact did the global pandemic have on banks?

Top 1000 World Banks ranking

Request a demonstration to The Banker Database

Join our community

The Banker on Twitter