MW AI investment 'advice' is 50% more likely to pump you up - and trip you into costly blunders
By Mark Hulbert
AI fuels impulsive actions; a human 'defense coach' is your best bet to win the market's 'loser's game'
Artificial intelligence in the investment arena has made real financial advisers more important than ever. That's because the best course of action in many situations is to do nothing, and investors' use of AI biases them toward action over inaction. A good financial adviser will be able to counter this bias.
There are three reasons why AI in the investment area leads to this bias. The first is a nearly universal tendency of human nature (referred to by psychologists as "action bias") to favor action over inaction, even when the evidence suggests that inaction would be better. Because of this bias, investors typically don't turn to AI when they think inaction is the right course of action.
A second reason is that large language models are trained on human behavior, so they tend to replicate the same biases.
Consider what Philip Resnik, a professor at the University of Maryland, wrote in the September 2025 issue of the journal "Computational Linguistics": "Harmful biases are thoroughly baked into what LLMs are. There is no bug to be fixed here. The problem cannot be avoided in large language models as they are currently conceived, precisely because they are large language models."
The third reason AI worsens investors' latent action bias: AI models tend to overly agree with users. So if an investor asks an AI chatbot if they should sell their stocks because war has broken out in the Middle East and oil's price has soared, it is likely to say that it's a good idea - even when it isn't. In other words, AI exacerbates the already-existing psychology tendency toward action over inaction.
AI's tendency to overly agree with users was confirmed by a study published recently in the journal "Science." The authors tested 11 leading LLMs, and found that they were 50% more sycophantic than human beings - defined as offering advice that is "flattering, people-pleasing, [and] affirming." Myra Cheng, the study's lead author and a computer-science PhD candidate at Stanford University, said that "by default, AI advice does not tell people that they're wrong."
Winning the 'loser's game'
Exacerbating investors' action bias wouldn't be so dangerous to your wealth if investing was a pursuit in which doing something - anything - is better than doing nothing. But most often this is not the case.
One of the most compelling arguments for why doing nothing works comes from a classic book written four decades ago by Charles Ellis, founder of the investment firm Greenwich Associates. Entitled "Winning the Loser's Game," the book described two types of contests - winner's games and loser's games - that differ according to whether risk-taking is necessary to win.
The distinction between these two types of games came from tennis. In a winner's game of tennis, the victor wins by making better shots. In a loser's game, the victor wins because of the opponent's unforced errors. Tennis for those of us who are not professional players is a loser's game, and we can win simply by lobbing the ball back to our opponents until they make unforced errors.
Ellis's insight is that investing is also a loser's game: We win by making the fewest unforced errors. We need to cultivate the investment equivalent of lobbing the ball back to the other side, which Ellis argued in most cases will be buying and holding a broad market index fund.
Mark Hulbert is a regular contributor to MarketWatch. His Hulbert Ratings tracks investment newsletters that pay a flat fee to be audited. He can be reached at mark@hulbertratings.com
More: No do-overs: How one extra dollar on your Roth conversion triggers a tax bill you won't see coming
Also read: Here's the next AI 'battleground' - and how investors can get in on the action
-Mark Hulbert
This content was created by MarketWatch, which is operated by Dow Jones & Co. MarketWatch is published independently from Dow Jones Newswires and The Wall Street Journal.
(END) Dow Jones Newswires
May 16, 2026 13:28 ET (17:28 GMT)
Copyright (c) 2026 Dow Jones & Company, Inc.
Comments