Business Tech

Americans believe financial tech tools are biased

A wide swath of the American public remains skeptical that the computer algorithms used to determine everything from Netflix recommendations to asset allocations remain free from human-like bias, according to a recent Pew Research Center survey. Majorities termed algorithm use for predicting criminal recidivism (56 percent) and resume screening (57 percent) but saved their greatest opposition (68 percent) for platforms that synthesize financial data into a “personal finance score.”

Several banks now offer customers “financial health” or “personal finance score” calculators — savings and investment apps provide similar measures — however, none have reported using the results from these public-facing apps in decision-making. Financial services professionals, too, have expressed concern over the encroachment of automated tools called “robo-advisors,” into traditional wealth management areas such as asset allocation, with many telling “Advisors Magazine” that human relationships are needed to truly understand investor needs.

The public’s reaction (and the industry’s, for that matter) might be overblown, however, given what these tools actually do.

“It’s not really accurate to conflate the automation of workflow with the automation of human judgement,” said Lex Sokolin, global director fintech strategy and partner at financial research firm Autonomous. Sokolin, previous to his work with Autonomous, developed and sold an automated financial advising platform.

“Robo-advisors take the very same investment analysis tools used by advisors and run customer data through the same mathematics that humans are using,” he told “Advisors Magazine” by email. “There is no noble end in having human operations staff doing the work of entering handwritten answers on paper into computer terminals — this can be done better through web, mobile, or speech interfaces.”

Unfairness and privacy topped the public’s list of concerns in the Pew survey. Researchers, however, disagree over how conflicted the algorithms are. A team of economists in 2002 studied an automated mortgage lending tool and found “that this increased accuracy results in higher borrower approval rates, especially for underserved applicants.” Other experts have disputed whether algorithms can be free of human bias.

The Pew survey found that demographic groups view algorithmic decision-making differently.

“Blacks and Hispanics are more likely than whites to find the consumer finance score concept fair to consumers. Just 25 [percent] of whites think this type of program would be fair to consumers, but that share rises to 45 [percent] among blacks and 47 [percent] among Hispanics,” the survey reported, possibly reflecting skepticism of financial services from traditionally underserved groups.

The financial world has become too complex to be solely managed by humans and the algorithms are not going anywhere, Sokolin said. This is especially true as human expertise remains expensive and prospective advisors seek out lower-fee alternatives.

“Think about trying to manage the millions of videos on YouTube — an approximation of human judgement needs to police these systems … Human intervention is needed across these landscapes, but it simply cannot be the default given the scale involved,” he said. “Further, most consumers don’t want to pay for the benefits of a human hand, while feeling entitled to receiving them.”


Follow Us

Subscribe to Our Newsletter

What's Next, Updates & Editorial Picks In Your Inbox

Related Articles

© 2017-2021 Advisors Magazine. All Rights Reserved.Design & Development by The Web Empire