Related Experiment Video
Updated: Jun 28, 2026

Virtual Agent for Real-Time Motivational Interviewing by Integrating Adaptive Nonverbal Behavior and Language Models
Published on: December 23, 2025
The Avatar Advantage: AI-Powered Faculty Development to Enhance Feedback Delivery Skills
Kara Michalsen1, Tai Lockspeiser, Angela S Czaja
1Ms. Michalsen: Coordinator, Academy of Medical Educators, Office of Medical Education, University of Colorado School of Medicine, Aurora, CO. Dr. Lockspeiser: Associate Dean of Assessment, Evaluation, & Outcomes, Office of Medical Education; Professor, Department of Pediatrics, University of Colorado School of Medicine, Aurora, CO. Dr. Czaja: Professor & Director of Faculty Development, Department of Pediatrics, University of Colorado School of Medicine, Aurora, CO. Mr. Carlson: Senior Instructor & Director of Education Technology, Department of Psychiatry, University of Colorado School of Medicine, Aurora, CO. Dr. Bilyeu: Assistant Professor, Department Head, Physical Therapy, College of Rehabilitation Sciences, University of Manitoba, Winnipeg, Manitoba. Dr. Kiger: Associate Professor and Director of Medical Education REsearch & Scholarship, Department of Pediatrics, University of Colorado School of Medicine, Aurora, CO. Dr. Gardner: Associate Dean for Faculty Development, Office for Faculty; Director, Academy of Medical Educators, Office of Medical Education; Professor, Department of Surgery. University of Colorado School of Medicine, Aurora, CO.
Background:
Feedback is fundamental to health professions education and essential for workplace-based assessment. Unfortunately, most educators lack formal training and access to evidence-based programs that incorporate experiential learning and real-time interactions.
Objective:
This study investigates the extent to which feedback delivery ratings provided by artificial intelligence (AI)-powered avatars are considered acceptable and reliable in comparison with feedback delivery ratings created by faculty participants and peer observers.
Methods:
After an interactive lecture on high-quality feedback, participants worked in small, facilitated four-person groups to deliver feedback in structured scenarios using AI-powered avatars. After each scenario, participants, peers, and the avatar independently rated feedback delivery using the DOCS-FBS scale (1-3; 1 = not done, 3 = successfully done). Scenario complexity was assessed with a nine-item scale (1-4; 4 = most complex). Multisource feedback ratings, postsession surveys, and group debrief data were analyzed with descriptive statistics, paired-sample t tests, and analysis of variance.
Results:
Participants rated session quality as very good (38%) or excellent (62%). Most found avatar-based feedback practice very or extremely effective (82%) and expressed strong interest in future avatar training (80%) and postworkshop access (74%). Participants reported high engagement and valued detailed rationales from the avatar. Peer observer ratings (2.82 ± 0.08) were significantly higher than both self-ratings (2.53 ± 0.39, P < .001) and avatar ratings (2.62 ± 0.32, P < .001). No differences emerged between self and avatar feedback delivery ratings.
Conclusions:
Using AI-powered avatars in feedback training sessions provide reliable feedback delivery ratings, are well received by participants, and can serve as a pragmatic solution for those leading professional development efforts.
Related Concept Videos
Impression Management Techniques IV: Altercasting
Impression Management Techniques III: Aligning Actions
Self-Evaluation: Self-Enhancement and Self-Verification
The Sense of Self: Reflected Self-Appraisal and Social Comparison