Evaluating Transparency Interventions to Reduce AI Over-Trust Among Vulnerable Populations

The increasing availability of general-purpose artificial intelligence (AI) tools, such as ChatGPT, has led many emerging adults (18–25 years) to use these systems for everyday personal concerns. Despite widespread adoption, this practice remains largely unregulated and has outpaced empirical research, raising questions about how users understand and rely on AI-generated information in personal contexts. This project examines how individuals interact with AI when discussing personal topics, such as stress and wellbeing. Specifically, it investigates factors associated with greater reliance on AI-generated responses and whether increasing transparency about how the AI generates its responses (e.g., by showing elements of its reasoning process) promotes more critical and informed use. The findings will contribute to the responsible design of AI systems, particularly for sensitive contexts where users may be more vulnerable and trust in AI-generated information is especially important.