Acceptability of AI-Assisted Psychotherapy to Prospective Patients and Their Intent to Continue Treatment: A Vignette Study.

Abstract

Objective: The authors examined how transparency of artificial intelligence (AI) use (transparent vs. covert) and modality of therapy (telehealth vs. text based) influenced prospective patients' perceived acceptability of AI assistance and intent to continue psychotherapy. A secondary objective was to assess whether age, AI familiarity, and prior psychotherapy experience predicted these perceptions.

Methods: A preregistered, cross-sectional vignette study was conducted with 201 U.S. adults recruited from an online survey platform. Participants evaluated five psychotherapy scenarios varying by modality and disclosure, including a control scenario with no AI use. After each vignette, participants rated acceptability and intent to continue treatment. Linear mixed-effects models were used to assess vignette effects, participant characteristics, and interactions involving age and AI familiarity.

Results: Across all models, therapist use of AI substantially reduced perceived acceptability and continuation intent compared with the control scenario (all p<0.001). Covert use consistently produced the strongest declines. Participant characteristics did not attenuate vignette effects. Older age and greater chatbot familiarity predicted slightly higher ratings and moderated transparent conditions, reducing penalties modestly.

Conclusions: Prospective patients showed strong anticipated preferences for psychotherapy without therapist use of AI. Transparency reduced, but did not eliminate, negative reactions. The findings suggest that potential implementation of AI-assisted psychotherapy requires explicit consent and careful attention to therapeutic relationship concerns.