“The committee was most focused on whether there are unintentional, bad results based on some biases that might be built into the data,” said Berman. “Morningstar’s panelist discussed about how over the years they’ve gotten better and better at making the algorithm better, using a committee that reviews it for fairness. So these companies recognize that there is a potential danger here.”
No Regulation – Yet
The meeting was preceded by remarks from SEC Chair Gary Gensler, who, in addition to the bias issue, voiced concern about artificial intelligence being used to generate behavioral nudges and discussed the potential conflict of interest in robo-advisors operated by asset managers trying to optimize their own revenue.
"Today, platforms have an insatiable appetite for data," said Gensler. "The underlying data used in the analytic models could reflect historical biases, or may be proxies for protected characteristics, like race and gender."
Yet Berman said there was no discussion in the meeting itself of behavioral nudges or an actual fiduciary or best-interest duty, and little discussion of regulation.
Regulation is unlikely unless there is evidence that shows a robo-advisor’s recommendation for one group of people is far worse than the recommendation for another group of people, and it turns out the algorithm is doing it, Berman said
“I think robo-advice is here to stay and it’s only going to grow. The big question will be whether they come out with regulations,” said Berman. “Then we have to ask whether those regulations will harm or help.”