AI Financial Advice Is Surprisingly Good, MIT Sloan Study Finds

About half of Americans say they use artificial intelligence for financial advice, yet little was known about the quality of that guidance. A new study by MIT Sloan researchers measures it directly. The paper, "AI Financial Advice: Supply, Demand, and Life Cycle Implications," was written by Taha Choukhmane, Weidong Lin and Matthew Akuzawa of MIT Sloan and Tim de Silva of Stanford's Graduate School of Business, and won the Swiss Finance Institute Outstanding Paper Award 2026.

How the study worked

The researchers first built a model describing how income, jobs, investments and taxes typically evolve over a lifetime, giving them a benchmark for good financial decisions. They then asked a sample of 1,000 adults to write their own prompts seeking spending and investing advice from GPT-5.2, GPT-5.6 or Gemini 3 Flash, and simulated what would happen if people aged 22 to 89 followed that advice repeatedly. Finally, they repeated the exercise with detailed academic prompts that included age, job status, income and savings balances, plus explicit assumptions about the economy.

Good advice, missed nuances

"We were somewhat surprised by how good the advice was," Choukhmane said. The models steered users toward higher savings during working years, drawing down savings in retirement, diversified stock funds and reduced stock exposure after age 45. But they relied on simple rules of thumb and adapted poorly to shocks: after a job loss, the AI urged spending cuts that were too sharp even for people with savings. The models also let portfolios drift instead of actively rebalancing them. Structured academic prompts improved the results.

Different prompts, different wealth

Advice varied with who was asking. Following responses to prompts written by men, financially literate users or people with prior AI experience produced about 5% more wealth near retirement. Prompts from women and less financially literate users led to roughly $50,000 (4%) less wealth at age 60, and those without earlier AI experience ended up almost $100,000 (6%) poorer. About two-thirds of the gender gap came from how people phrased their questions — men favoured words like "strategy" and "growth", women words like "family" and "pay" — while a third came from the model changing its advice for otherwise identical prompts.

What it means for consumers

The study also found LLMs recommending products users never mentioned: Vanguard appeared in 6% of responses and iShares in 3.4%, although under 0.4% of prompts named either. For consumers, the lesson is to prompt like a planner — grounding questions in life-cycle planning, portfolio theory and real-world assumptions. Choukhmane suggests using AI first to build financial understanding, and as an inexpensive complement to a human adviser, especially for people who cannot afford one.