FinovateEurope: Nasdaq’s behavioral sciences head says AI can be “nonsense”
Nasdaq’s head of behavioural sciences for market technology risk and surveillance says some uses of artificial intelligence (AI) can be “nonsense”.
On a panel about financial crime at FinovateEurope in Berlin, Wendy Jephson discussed the limitations of using just one technology such as AI to solve universal problems such as money laundering, insider trading and market manipulation.
Wendy Jephson at FinovateEurope
“People talk about AI as if it can solve anything, […] but it isn’t the only answer,” she says. “We need to think hard about the problems we’re solving and what technology is the answer. For that, you need a whole tool kit”.
Jephson reminisces on talking to someone in the industry who worked in deep behavioral machine learning, and he’d said he was using “emotional recognition” to detect “deception”. She concluded: “You can’t detect deception from emotional recognition, that’s nonsense”.
Oxagile, a New York-based emotion recognition solution driven by machine learning (ML), promises fraud detection as one of its capabilities in financial institutions.
The skeptical Nasdaq head, who worked in a family-owned pharmaceutical company for 13 years, is also an advisory council member for non-profit regulation alliance AIR.
“‘We want to stop financial crime’ is half of the sentence,” says Jephson. “The other half is about running the financial institution efficiently at the same time.”
More from Finovate Europe: RBS head says bank has “turned a corner” in its innovation culture
She cites a conversation with Nick Leeson, the former English derivatives broker who went rogue and brought down the UK’s oldest merchant bank Barings into bankruptcy in 1995.
When Leeson was working at Morgan Stanley, he told Jephson that if your positions – the market commitments traders make – did not match then you could no longer trade until they did match. The trader either had to figure it out and make them match, or call someone “scary” in London and own up to the oversight. Leeson said he figured it out in the end and made the positions match.
Whilst Jephson does not refute that AI can be very helpful – namely in the effort of speed, something she feels the industry has the least of – she also reminds the Finovate audience that what are now deemed to be traditional methods should not be shunned or forgotten.
“These people [insider criminals] are opportunistic,” she says. “They are always changing and adapting which is why speed is important.”
But just as AI is great contender to speed up processes, Jephson concludes by reminding the Finovate audience that just one cubit centimeter in all our brains holds more connections than stars in an entire milky way. The lesson being, do not underestimate the timeless power of the brain amidst the emergence of new technologies.
