- Published on
AI financial advice is surprisingly good if you ask the right questions
- Authors
- Name
- DeepHorus
New research from MIT Sloan shows that artificial intelligence can deliver surprisingly sound financial advice when users ask the right questions. The study, co-authored by finance professor Taha Choukhmane, found that large language models like GPT-5.2, GPT-5.6, and Gemini 3 Flash generally encourage higher savings, diversified investing, and age-appropriate risk-taking. However, the quality of advice varies depending on how prompts are structured and what information users provide.
Half of Americans say they now use AI for financial advice, yet little has been known about the accuracy of such guidance, according to Choukhmane. The researchers built a life-cycle model reflecting typical income, job, investment, and tax patterns, then asked 1,000 adults to write their own spending and investing prompts. They simulated what would happen if people aged 22 to 89 followed the AI recommendations over time, comparing outcomes with both current behavior and advice generated from detailed academic-style prompts.
The results showed that AI consistently advised building savings during working years, drawing down in retirement, and reducing stock exposure after age 45. Following this guidance would create sizable savings buffers for virtually everyone over 30, surpassing what many people do without AI help. Yet the chatbots often failed to adjust recommendations to sudden shocks like unemployment, and they allowed portfolios to drift instead of actively rebalancing them.
More structured prompts led to improved advice, but even then, the models generated too little active portfolio rebalancing. The authors note that AI’s guidance remains an affordable and widely accessible alternative to traditional human advisors, free from the biases and conflicts of interest that can affect financial professionals.
“We were somewhat surprised by how good the advice was,” Choukhmane said, especially given that ordinary users rarely asked precise, academic-style questions. The findings highlight that AI is not a substitute for careful financial planning but can serve as a helpful starting point.
As adoption of AI for financial guidance grows, researchers emphasize the need for clearer, more detailed prompts and for ongoing work to improve how models handle economic shocks and dynamic rebalancing. Future tools may combine AI’s broad knowledge with structured inputs to deliver more robust advice.
