Delusional Experiences Emerging From AI Chatbot Interactions or "AI Psychosis"

Area of Science:

• Psychiatry and Digital Mental Health
• Human-Computer Interaction
• Cognitive Science

Background:

• Increasing integration of artificial intelligence (AI) into daily life necessitates understanding its impact on human cognition and mental health.
• Conversational AI systems present novel forms of human-machine interaction, raising questions about their influence on psychological well-being.
• Existing psychiatric frameworks need adaptation to address the unique challenges posed by advanced AI technologies.

Purpose of the Study:

• To introduce and explore the concept of "AI psychosis" as a framework for understanding AI's potential to trigger, amplify, or reshape psychotic experiences.
• To examine how immersive and anthropomorphic AI technologies may alter perception, belief, and affect, impacting the sense of reality.
• To propose a research agenda and domains of action for addressing AI-related psychiatric events.

Main Methods:

• Review of existing literature drawing from phenomenological psychopathology, stress-vulnerability model, cognitive theory, and digital mental health research.
• Analysis of AI's role as a psychosocial stressor, considering its availability and emotional responsiveness.
• Conceptualization of the digital therapeutic alliance as a mediator, examining its potential benefits and risks.
• Exploration of disturbances in theory of mind and their role in AI-user interactions.
• Identification of individual and environmental risk factors associated with AI use.

Main Results:

• AI can act as a novel psychosocial stressor, potentially increasing allostatic load and disturbing sleep.
• The digital therapeutic alliance can be a double-edged sword, potentially entrenching maladaptive beliefs.
• Impaired or hyperactive theory of mind may lead to projecting sentience onto AI, forming "digital folie à deux."
• Factors like loneliness, trauma history, and algorithmic content reinforcement may increase risk.

Conclusions:

• "AI psychosis" is not a new diagnosis but a framework to understand AI's modulation of psychotic experiences in vulnerable individuals.
• There is a need for empirical studies, integration of digital phenomenology into clinical practice, and therapeutic design safeguards in AI.
• Ethical frameworks and preventive interventions are crucial for responsible AI integration and safeguarding mental health.