Beyond the black box: why algorithms cannot replace the unconscious or the psychodynamic therapist

Area of Science:

• Psychoanalytic theory
• Clinical psychology
• Artificial intelligence studies

Background:

• The increasing integration of artificial intelligence (AI) into mental healthcare necessitates a critical evaluation of its capabilities.
• Existing AI models often struggle to capture the complexities of human psychological experience, particularly unconscious processes and therapeutic relationships.

Purpose of the Study:

• To critically examine the limitations of AI in replicating human psychological processes.
• To challenge AI's capacity to understand the unconscious and psychotherapeutic dynamics.
• To highlight the structural limitations of AI in psychological care.

Main Methods:

• Theoretical analysis drawing on psychoanalytic concepts (Matte-Blanco's unconscious logic, Winnicott's therapeutic holding).
• Presentation of a clinical vignette illustrating AI's therapeutic shortcomings.
• Analysis of AI's inability to replicate key therapeutic elements like silence, countertransference, and containment.

Main Results:

• AI fundamentally fails to engage with the non-linear, contradictory, and embodied nature of human psychology.
• AI cannot replicate intersubjective, temporal, and affective dimensions crucial for psychological understanding.
• A significant risk identified is patients altering self-presentation to fit computational logic.

Conclusions:

• AI's limitations in psychological care are structural, not merely technical.
• Embodied, relational human connection remains irreplaceable in psychological understanding and treatment.
• AI may serve supplementary roles but cannot substitute for human therapeutic presence.