Are you qualified, ChatGPT? Examining clinical skills and competencies of ChatGPT in delivering systemic interventions

Abstract

Large language model agents such as ChatGPT are increasingly used for relationship advice and mental health support, with many users describing these interactions as therapeutic. This study assessed the clinical and cultural competence of a GPT‐based model trained to provide systemic interventions aligned with the AAMFT Code of Ethics and core skills. A volunteer client discussed a relational concern with the model, and the resulting transcript was evaluated by 12 experienced supervisors using standardized measures. Supervisors, unaware of the study's purpose and the therapist being an AI, rated the model as moderately competent in maintaining a systemic perspective and delivering systemic interventions. Supervisors identified weaknesses in pacing, joining, and attending to cultural and contextual factors, and inferred that the therapist resembled an early‐career clinician with less than 500 h of experience. Findings indicate limitations of AI therapy and underscore the value of human judgment, attunement, and intuition in clinical practice.