AI financial advice is surprisingly good, especially if you ask right questions
Points and comments are a snapshot, not live.
AI financial advice encourages saving and diversified investing, but misses nuanced planning.
A new MIT Sloan paper, which won the Swiss Finance Institute Outstanding Paper Award 2026, tests AI financial advice from GPT-5.2, GPT-5.6, and Gemini 3 Flash. Researchers simulated life-cycle outcomes for 1,000 adults using their own prompts and compared them to advice from structured academic prompts. AI consistently advised higher savings, diversified stock funds, and reduced stock exposure after age 45. However, it failed to adjust for shocks like unemployment and allowed portfolios to drift. Advice quality improved with more detailed prompts, but wealth gaps emerged based on gender, financial literacy, and AI experience-leading to up to $100,000 less in wealth at age 60 for some groups.
The study found that LLMs often recommended specific financial products (e.g., Vanguard in 6% of responses) even when users did not mention them, suggesting AI may reshape how consumers discover financial products. The authors note that AI advice is a good complement to human advisors, especially for those with limited resources.
What commenters are saying
Commenters generally agree AI financial advice is decent but generic, echoing common principles like saving more and diversifying. Many point out that real value lies in behavioral coaching, nuanced tax advice, and personalized planning-areas where AI often falls short. Several commenters share practical experiences: one exports YNAB data to Claude for tailored budgeting and tax insights, while another warns that LLMs miss local tax rules (e.g., NYC S Corp taxes). A split emerges between those who see AI as a useful tool for data analysis and those who argue it cannot replace human advisors who ask clarifying questions and provide emotional support. Some note that AI advice can vary significantly by user demographics, reinforcing wealth gaps.