Minutes

Mohammad Amin Kuhail, Nazik Alturki, Justin Thomas, Amal K. Alkhalifa & Amal Alshardan (2025). Human-human vs human-AI therapy: An empirical study. International Journal of Human–Computer Interaction, 41(11), 6841–6852. https://doi.org/10.1080/10447318.2024.2385001

Could an AI chatbot provide therapy that feels as helpful as talking with a human therapist? United Arab Emirates and Saudi Arabian researchers examined whether mental health professionals can tell the difference between human-led and AI-generated therapy conversations. 

The study included 63 therapists, including psychologists, counselors, and psychiatrists. Participants read therapy transcripts from either traditional therapist-client interactions or conversations between clients and an AI chatbot called Pi. They then rated the quality of each session and try tried to identify whether it comes came from a human or an AI. 

The findings are surprising. Therapists correctly identified the source of the transcripts only 543.9% of the time - no better than chance. They also rate the AI-generated conversations as higher quality on average than the human-generated transcripts. 

This matters because many communities face shortages of mental health professionals and long waits for care. The results suggest that AI chatbots may be able to provide supportive listening and initial guidance when therapists are unavailable. 

Conclusions? AI cannot replace human therapists, but it may help expand access to mental health support if it is monitored. As these tools improve, they could become a valuable first step for people seeking help. 

Written by Dan Peacock B.S.