The influence of AI-powered investment recommendations on individual investors: A mixed-methods study from the perspective of persuasive communication

This study examines how individual retail investors receive AI-generated investment advice from the perspective of persuasive communication. Arguing that large language models (LLMs) create a persuasion dynamic different from that of traditional media, it combines in-depth interviews with a content analysis of AI outputs in a mixed-methods design. Interviews with 18 investors who use AI for investment decisions were analyzed through Interpretative Phenomenological Analysis, and the responses of four models to six standardized scenarios were coded with the AI Investment Advice Evaluation Scale (AIIAES), a researcher-developed instrument of eight dimensions and 24 items. The findings show how framing, the machine heuristic and sycophancy shape user decisions, and they indicate that financial literacy differentiates responses without protecting users evenly. No model produced a complete disclaimer in any scenario, and Gemini displayed a scenario-dependent sycophancy pattern. On the theoretical side, the study describes a deliberate form of user-induced sycophancy, in which users steer a model over repeated interactions toward confirming their own preferences. On the methodological side, it offers an instrument that turns the suitability and disclosure standards of investment regulation into criteria for conversational AI outputs. The study concludes with policy recommendations for user protection.