AI Financial Advice: Supply, Demand, and Life Cycle Implications

Abstract

As more households turn to AI tools for financial guidance, what advice do they receive? The answer depends on the model used (supply), on the prompts people write (demand), and on their evolving circumstances over the life cycle. We survey a representative sample of adults, have them write prompts seeking spending and investing advice, and simulate the lifetime paths of following that advice under realistic asset and labor market conditions. Applying our method to GPT-5.2 and Gemini 3 Flash, we document three facts. First, following this advice would move most respondents toward life cycle theory: broader equity participation, equity shares that decline with age, and larger savings buffers. Second, replacing individual prompts with more structured ones moves advice closer still, improving consumption smoothing and reducing reliance on simple heuristics. Third, advice varies systematically with characteristics like gender, prior experience using AI, and financial literacy. For gender, two-thirds of the difference in recommended equity shares reflects demand (what men and women ask) and one-third supply (how models respond to identical prompts labeled by gender). These findings suggest that generative AI has the potential to improve financial decision-making, but that its impact is likely heterogeneous across households.