AI financial advice is surprisingly good, especially if you ask right questions

Artificial intelligence is becoming a popular source of financial advice, but the question remains whether following that advice will improve people's financial standing. According to Taha Choukhmane, an MIT Sloan School of Management professor, half of Americans are using AI for financial advice, but there's limited understanding of the type of advice provided and whether it's acted upon.

Choukhmane and his co-authors found that following AI recommendations can lead to significant savings for individuals over 30. AI consistently advised people to save during their working years, reduce investments after age 45, and draw down savings during retirement. However, AI struggled to adapt to life changes like unemployment and didn't actively rebalance portfolios.

The quality of advice improved with more structured prompts, but AI still generated insufficient active rebalancing. Researchers built a model reflecting typical life events such as income, jobs, investments, and taxes, and then asked 1,000 adults to write prompts seeking spending and investing advice from GPT-5.2, GPT-5.6, or Gemini 3 Flash.

They simulated the outcomes for individuals aged 22 to 89, comparing AI advice to actual financial behavior and academic prompts with full financial information. The findings showed that LLMs offer affordable, accessible financial guidance that can help overcome traditional advisors' costs, biases, and conflicts. The AI advice was better than expected, promoting higher savings, increased stock market participation, and diversified allocations.

However, AI advice falls short on nuances, relying on simple rules of thumb and failing to adjust to changing circumstances. The way people ask questions also affects the advice, with more detailed prompts yielding better results. Additionally, the advice varies based on the user's gender, financial literacy, and experience, leading to potential wealth gaps.