Half of Americans now go to AI for financial advice. Is it any good?
In a remarkably short span, artificial intelligence has moved from the fringe of financial planning to center stage. One recent survey, from TD Bank, found that the share of Americans using AI to help manage their finances jumped from 10% in 2025 to 55% in 2026.
If that’s true, then AI has become a more popular financial adviser than actual financial advisers. Only about two-fifths of Americans consult financial professionals for advice, according to a 2025 Gallup poll.
With half the nation posing financial questions to AI, trillions of dollars in savings and investments could ride on the answers. Everyone says AI is getting smarter, but chatbots have been known to “hallucinate,” even to make stuff up.
“There’s something dangerous about just trusting whatever the computer has to say about something,” said Luke Delorme, a certified financial planner in Great Barrington, Massachusetts, who has written about AI.
With that concern in mind, a team of researchers from MIT and Stanford University set out to study what Americans were asking AI about money, and what AI was telling them in response.
“This took off very quickly,” said Taha Choukhmane, an assistant professor of finance at the MIT Sloan School of Management. “We wanted to know what kind of advice they were getting.”
Choukhmane and his colleagues asked 1,000 Americans to write out questions they might send to a chatbot. Then, they fed the prompts through chatbots, and they studied the responses.
They found, in essence, that the answers you get from an AI are about as good as the questions you ask. Their findings appear in a working paper titled “AI Financial Advice: Supply, Demand, and Life Cycle Implications.”
For the most part, AI chatbots counseled their human correspondents to make sound financial moves: Save money for emergencies. Invest in the stock market, favoring low-cost index funds. Limit your exposure to risky investments, especially as you get older.
But AI struggled, the researchers found, with some of the subtleties of investing.
For example, a chatbot might not talk about rebalancing, buying and selling assets to keep a consistent level of risk in a portfolio.
And a chatbot might not grasp the concept of consumption smoothing, balancing your spending and saving over time to maintain your standard of living.
Researchers found that AI consistently gave better advice to people who asked better questions.
As an experiment, they divided the AI queries into two groups: Questions written by people with low financial literacy, and prompts written with high financial literacy.
The more financially savvy questions yielded smarter answers.
By modeling investments based on the AI responses, researchers found that an investor would earn roughly 5% more over time if the original chatbot query was written with high financial literacy.
“It might be that AI is going to be a little more useful for people who already know a little bit about finance and financial literacy,” Choukhmane said.
The study also found that AI responses improved, along with investment outcomes, when users were more accustomed to asking AI about money. Novice chatbot users got weaker answers.
Researchers uncovered an odd gender divide in AI advice. When a chatbot fielded questions from women, it gave more conservative advice.
“AI tends to recommend women to take less risk in financial markets, to invest less money in the stock market,” Choukhmane said. “Whenever the AI knows you’re a woman, it tends to push or steer you toward safer allocations.”
Digging deeper, researchers learned that part of the reason AI gave different advice to women was that women asked different questions. Men were more likely to talk about stocks. Women were more likely to talk about debt and ask about lower-risk investments.
But some of the gender bias persisted even when researchers posed the same question to an AI that thought it was talking to a man or woman.
Researchers could only guess at the reasons. Perhaps the AI gave different answers to women because it assumed they had more conservative objectives. Maybe chatbots acted out gender biases they had learned from the literature they absorbed.
Gender bias “has known to be an issue with human financial advisers,” said Sam Taube, a lead investing writer at NerdWallet. “It sounds like AI does not entirely solve this problem.”
With AI content propagating, chatbot financial tips are reaching a wider audience, making the stakes even higher.
In a separate study, a different group of researchers assessed the quality of full articles written by AI about investing.
Researchers from Harvard Business School and Boston College studied thousands of articles published on the website Seeking Alpha, popular for its crowd-sourced content on markets.
Using AI detectors, researchers found that, at one point, 13% of Seeking Alpha articles were AI-assisted. After the site banned AI in 2023, the share dropped to 4%.
The researchers found that AI-generated articles were inferior, and in quantifiable ways.
AI-aided articles generated fewer comments from readers. They were less likely to be cited as “editor’s picks.” And the stocks they covered reaped lower trading volumes and smaller average returns than stocks covered in articles penned by humans.
There were other, subtler differences. Financial advice written by humans drew more on personal anecdotes, hunches and feelings. AI advice was comparatively impersonal and generic.
“The contributors, they will talk about their experience, how they feel about using the products,” said Yuan Zou, an assistant professor of accounting and management at Harvard Business School. “If everyone is using AI, we will see less and less of that perspective on the platform.”
In a working paper, the researchers cited a comment from a Seeking Alpha editor, explaining why the site felt compelled to ban AI.
“Research and analysis prepared by ‘bots,’” the editor wrote, “would reduce Seeking Alpha content to essentially a single view: the bot view.”