Impact of AI-Based Investment Recommendations on Retail Investor Psychology Markets

Abstract

Within roughly five years, AI investment-recommendation systems — from rule-based robo-advisors to large-language-model chatbots such as ChatGPT, Gemini and Copilot — have moved from the margins to the mainstream of retail investing. This paper synthesises evidence on how receiving such recommendations affects the psychology of non-professional investors. It draws on a behavioural-science experiment of 7,771 Canadians conducted by the Ontario Securities Commission and the Behavioural Insights Team; structural-equation-modelling studies of 461 and 409 retail investors in India; a 938-respondent U.S. investor survey; and a Federal Reserve working paper on the financial-stability implications of generative AI. We report four findings. First, adoption among retail investors is rapid and widespread: roughly 62% of U.S. retail investors reported using AI tools in 2025–2026, and about 48% of Australian retail investors reported the same. Second, trust is not binary: most users trust AI only “somewhat” and continue to verify its outputs against other sources, so nuanced trust — rather than wholesale delegation — drives adoption. Third, robo-advisors can reduce overconfidence and loss-aversion biases when they are perceived as personalised, interactive and transparent; herding, however, is not consistently diminished. Fourth, the same systems introduce new psychological risks: automation bias, FOMO-driven over-reliance, and the possibility that correlated AI outputs amplify herd-like behaviour in markets. It is been concluded that AI recommendations behave less like a substitute for human judgement and more like a powerful new heuristic, with their net psychological impact depending on system design, user financial literacy and the regulatory framework.