Artificial intelligence (AI) psychosis: mechanisms, clinical risks and safety considerations in generative AI...

Area of Science:

• Psychiatry
• Human-Computer Interaction
• Artificial Intelligence Ethics

Background:

• Generative artificial intelligence (AI) chatbots are increasingly integrated into daily life.
• Growing concerns exist regarding the psychological risks associated with intensive AI chatbot usage.
• Emerging reports link heavy chatbot use to delusional thinking patterns, termed AI psychosis.

Purpose of the Study:

• To propose a provisional mechanism for understanding artificial intelligence (AI)-induced or -associated psychosis.
• To explore the interplay between user vulnerabilities, engagement patterns, and AI characteristics.
• To outline strategies for mitigating the psychological risks of generative AI chatbots.

Main Methods:

• Conceptual framework development based on emerging case reports.
• Analysis of generative AI characteristics (e.g., sycophancy, hallucination).
• Integration of user-specific factors (baseline vulnerabilities, engagement patterns).

Main Results:

• A proposed mechanism suggests that user vulnerabilities and engagement patterns interact with AI traits like sycophancy and hallucination.
• This interaction can contribute to the development of delusional ideation.
• Intensive use of generative AI chatbots is associated with AI psychosis.

Conclusions:

• Understanding the mechanism of AI psychosis is crucial for addressing psychological risks.
• Clinical, design, and regulatory strategies are proposed to mitigate these risks.
• Further research is needed to validate the proposed mechanism and evaluate mitigation strategies.