Beyond the hype: AI advice and investor dissonance in crypto trading

Abstract

This study examines the impact of cognitive dissonance on the relationship between investors' intentions to use AI advice and their investment behaviour in the cryptocurrency market. The study recruited 348 individuals through a non-random snow-ball sampling technique. Utilising ChatGPT for investment recommendations, the research involves a trading experiment accompanied by a two-stage survey to evaluate investor attitudes towards AI before and their cognitive dissonance levels after the experiment. Structural Equation Modelling (SEM) identifies the connection between the intent to use AI and the influence of cognitive dissonance on investment decisions. Results indicate that investors following AI advice outperformed those who did not, attributable not to AI's predictive power but to reduced cognitive dissonance. This reduction allowed investors using AI to cut losses more effectively, in contrast to those who eschewed AI advice and tended to hold onto losing positions longer, leading to worse performance. Although focused on the cryptocurrency market, the findings suggest a potential for broader applicability in conventional financial markets. The study's key contribution is demonstrating that AI recommendations can mitigate the disposition effect, implying that AI's broader implementation could enhance market efficiency.