Banks Deploy IBM AI Software to Catch Errant Traders before Misconduct Occurs


Marc Wilson, head of the Financial Services division of IBM Watson said the tools that are capable of anything from monitoring conversations to detect signs of traders going rouge / Photo by: Atomic Taco via Wikimedia Commons


Marc Andrews, vice president of IBM’s Watson Financial Services division, outlined the division's tools that are capable of doing everything from monitoring conversations for a tone to using changes to credit scores in order to detect signs of traders likely going rogue.

Watson, IBM's AI software, has already been deployed by a dozen banks in monitoring their communications every day. Andrews said that the tool looks at the emails of bankers, as well as identifies their communication patterns, from the tone of their email up to what they are saying, the Financial Times reports.

“One of the benefits of…applying machine learning is that you’re not implementing specific rules that someone can just work their way around,” the division vice president said, adding: "As people do start changing their behavior...the models will learn over time and will be able to adapt much more quickly.”

Financial Times says an important outcome of this is the decline of "false positive" banks that have to face traditional systems that flag hundreds of thousands of possible suspect messages per month, which leave banks to find a needle in a haystack.

It adds that the Watson AI software varies between higher and lower-risk alerts in order to allow financial institutions to cut through to the number of messages they receive. Andrews said the idea is to cut down the effort and cost that are spent on low-risk incidents.

He added that banks are already including other metrics into their surveillance systems, which includes reviews of human resources and credit scoring reports in an effort to determine traders with the motive for going rogue.

Moreover, IBM has separately devised tools that would let other information to be integrated into the banker surveillance system. Red flags include a court conviction or a hefty amount of divorce settlement.

“We’ve had a lot of inquiries about that, but none have started putting it into production,” Andrews said. He further stated that a lot of banks are mainly in that experimental phase, as they try to identify the techniques that would be worth investing in.