AI financial advice is surprisingly good, especially if you ask right questions (mitsloan.mit.edu)

A new study co-authored by Taha Choukhmane from MIT Sloan reveals promising results regarding the quality of financial advice provided by large language models (LLMs). With half of Americans reportedly using AI for financial guidance, the research indicates that following AI recommendations can significantly enhance the financial wellbeing of individuals aged 30 and over. Key advice from LLMs includes encouraging savings during working years, prudent investment in diversified stock funds, and age-appropriate risk management. However, the models struggled with adapting to sudden financial setbacks and often failed to recommend necessary portfolio rebalancing. The significance of this research lies in its potential to democratize financial advice, particularly for those who may not afford traditional advisors. Notably, the quality of AI-generated advice improved with more structured prompts, highlighting the importance of how users frame their questions. Despite the advantages, disparities exist in the advice tailored to different demographic groups, leading to wealth gaps. By acknowledging the impact of user prompts—such as those influenced by gender or financial literacy—the study underscores the need for a nuanced approach to AI financial guidance that could address biases and ultimately enhance outcomes for all users.