AI Appreciation and Financial Advice

Generative artificial intelligence (AI) is about to transform financial advice by enabling tailored solutions for investors at low cost. However, an aversion to artificial intelligence and lack of trust in recommendations generated by AI models could prove to be a major obstacle to their broad introduction. We test AI aversion in the context of financial advice in three incentivized experiments (N=1,176). Participants receive investment recommendations sourced either from ChatGPT or from a financial professional. The rate at which participants follow the recommendations and their satisfaction with the advice is consistently higher in the AI treatments. Observing intermediate investment outcomes weakens AI appreciation as outcomes distract from recommendation quality. Participants do not anticipate their AI appreciation, as a majority selects the financial professional in an experiment with free advisor choice. This suggests uncertainty surrounding AI capabilities, which is only resolved when seeing the actual recommendations.

Download

JEL codes: D14, G11, G41, G50.

Keywords: Artificial intelligence, Algorithm aversion, Financial advice, Robo advice, Risk preference.