Symposium
Artificial Intelligence and Technology-based Interventions
Daniel Szoke, Ph.D. (he/him/his)
Rush University Medical Center
Chicago, Illinois, United States
Ilana Hutzler, BA (she/her/hers)
Research Assistant
Rush University Medical Center
Chicago, Illinois, United States
Sarah Pridgen, M.A.
Research Manager
Rush University Medical Center
Chicago, Illinois, United States
Katy Dondanville, PsyD, ABPP
Associate Professor
The University of Texas Health Science Center at San Antonio
San Antonio, TX, United States
David C. Rozek, ABPP, Ph.D.
Assistant Professor
University of Central Florida
Orlando, FL, United States
Philip Held, PhD
Associate Professor
Rush University
Chicago, Illinois, United States
Training clinicians in evidence-based treatments is often limited by cost and time constraints. This creates a need for novel training approaches. Generative AI may be leveraged to supplement established clinical training approaches by providing accessible, low-stakes opportunities to practice skills. This study evaluated clinician satisfaction with Socrates Coach, an AI platform that simulates patient conversations and provides line-by-line, as well as overall, feedback on the clinician’s Socratic dialogue techniques.
Clinicians with at least a master's degree (N=22, Mage=43.45, 90.91% female, 77.27% master's level) were recruited through the STRONG STAR Training Initiative and given unlimited access to Socrates Coach for one month. Participants completed a qualitative interview and surveys assessing usability (MAUQ), acceptability (AIM), appropriateness (IAM), and feasibility (FIM).
Usability scores were high (M=6.00, SD=0.66, range 1-7). Clinicians reported high satisfaction (M=6.36, SD=0.95, range 1-7) and usefulness for practice (M=6.41, SD=0.96, range 1-7). Ratings were also high for acceptability (M=4.42, SD=0.52, range 1-5), appropriateness (M=4.40, SD=0.63, range 1-5), and feasibility (M=4.46, SD=0.52, range 1-5). In qualitative interviews, clinicians affirmed that Socrates Coach supplemented their formal clinical training. Clinicians appreciated the immediate AI-generated feedback, which evaluated their use of Socratic dialogue. Clinicians noted that the tool felt less intimidating and more accessible than traditional supervision, as they could practice without fear of judgment.
Clinicians reported high satisfaction with Socrates Coach across all assessed domains. Although preliminary, findings suggest that AI-based training tools like Socrates Coach may complement clinical training by providing affordable, accessible, and real-time feedback that can enhance skills practice. AI-based training tools could be valuable for clinicians with limited supervision access or seeking additional practice. Future research should explore whether engagement improves Socratic dialogue skills through standardized competency assessments.