The study, published in the Journal of Financial Planning, analyzed seven generative AI platforms, including ChatGPT and Claude, and found notable inconsistencies in their financial advice.
Researchers noted that while AI can provide general insights, such as the importance of diversification, its outputs can be misleading or incorrect due to algorithmic 'hallucinations' and sensitivity to user prompts. Andrew Lo from MIT cautioned that AI's authoritative-sounding responses can be deceptive, particularly for specific financial calculations.
The research also indicated that demographic factors could influence the recommendations provided by these platforms, raising concerns about fairness and consistency.
With 66% of Americans using AI for financial advice, and even higher percentages among younger generations, the findings suggest that while AI can be a useful starting point for financial inquiries, it should not replace professional advice. The study underscores the evolving nature of AI tools and the potential differences in outputs between free and paid models