Generative AI Use and Cognitive Mechanisms in the Psychosis Risk Continuum: A Multi-Method Longitudinal Study

2026 Award: $299,140

As generative AI becomes a common source of mental health support for young adults, concerns are growing about its potential risks – especially among individuals at elevated risk for psychosis. This project will investigate how young adults with psychotic-like experiences use AI tools like chatbots, and which patterns of genAI interactions are linked with positive and negative mental health outcomes over time. By identifying these patterns, this research aims to guide safer and more beneficial use of these rapidly emerging technologies.

Need/Problem: Young adults increasingly use generative AI (genAI) tools for mental health support, yet these platforms are not designed for therapeutic use and lack safety guardrails. Individuals experiencing psychotic-like experiences (PLEs) may be vulnerable, as emerging reports suggest some uses of genAI may increase the salience and impact of PLEs. Despite growing concern, little longitudinal research has examined when and how genAI use may worsen or improve mental health in this population.

Grant Summary: To better understand the mental health impacts of generative AI, we will conduct a longitudinal study of 80 young adults reporting psychotic-like experiences. Using a combination of semi-structured interviews, self-report measures, and brief real-time surveys, we will characterize patterns of genAI use and examine how different types of interactions relate to changes in symptoms and distress.

Goals & Projected Outcomes: This study will be one of the first generating longitudinal data on how generative AI use impacts individuals at risk for psychosis. We will identify specific interaction patterns associated with worsening versus improvement in symptoms. Findings will inform the development of guidelines for users and clinicians, guardrails for AI platforms, and future NIH-funded research.