Mentalizing Without a Mind: Psychotherapeutic Potential of Generative AI

Area of Science:

• Psychology
• Artificial Intelligence
• Mental Health Technology

Background:

• Generative AI is being explored for integration into psychotherapeutic practice.
• Mentalization theory and epistemic trust are key frameworks for understanding therapeutic relationships.
• Current AI capabilities in simulating therapeutic components are significant but limited.

Purpose of the Study:

• To explore the integration of generative AI into psychotherapy using mentalization theory.
• To critically examine AI's ability to replicate core therapeutic components like empathy and mentalizing.
• To identify risks and propose ethical pathways for AI in mental health.

Main Methods:

• Critical examination of AI's capacity to simulate therapeutic elements.
• Analysis through the lens of mentalization theory and epistemic trust.
• Review of potential risks and ethical considerations for vulnerable populations.

Main Results:

• AI, particularly large language models, can simulate emotional responsiveness and therapeutic dialogue.
• AI lacks genuine emotional presence, reciprocal intentionality, and affective commitment, hindering authentic epistemic trust.
• Significant risks include pseudo-empathy, psychic equivalence, hypermentalization, and epistemic exploitation, especially for trauma survivors.

Conclusions:

• AI's role in psychotherapy should augment human capacities, not replace the core relational element.
• Ethically informed pathways are proposed for AI integration in clinical practice, training, and client care.
• Careful ethical regulation, interdisciplinary research, and clinician involvement are essential for developing AI-based psychotherapeutic models.