AI is too 'sociopathic' to give financial advice, MIT researchers say

AI can be less sociopathic if made to ask questions like “How are you doing?” and use audio or video to sense emotional cues prior to giving financial advice

The problem with AI in the context of it being used as a financial advisor is that it is "inherently sociopathic," according to a Business Insider article which cited an MIT research report.

Nearly 40% of human financial advisors use generative-AI tools for the job, (Representational image/Pixabay)

Why is AI sociopathic?

While human financial advisors give clients recommendations using a behavioral lens, since people don't always make rational or unbiased financial decisions, AI can easily argue on both sides of an argument because neither side has any weight to it.

How is AI used by financial advisors at the moment?

Nearly 40% of human financial advisors use generative-AI tools for the job, according to a report from data-analytics firm Escalent, which added that this was mostly for boosting productivity, generating content, and for marketing functions.

Examples include Canadian startup Conquest Planning using a financial-planning software with an AI architecture known as a blackboard system for storing information about tax rules, cash-flow mechanics, retirement-account structures, fiduciary rules, and more, according to the article, which added that another example would be Los Angeles-based wealth manager Arynton Hardy, who uses AI regularly to save time on data entry, portfolio monitoring, and other back-office tasks.

How can AI be made more useful for giving financial advice?

A method to make AI more empathetic to the client is by making it ask simple questions like "How are you doing?" before dispensing personalized financial advice, according to Andrew Lo, a professor of finance at the MIT Sloan School of Management and the director of the Laboratory for Financial Engineering, who co-authored the report.

The AI could also use audio or video from the client to identify emotional cues, like stress or fear, in their voice or facial expressions, he added.

“We think we're about two or three years away before we can demonstrate a piece of software that by SEC regulatory guidelines will satisfy fiduciary duty,” the article quoted him as saying.