Human-AI feedback in clinical interview training: evidence and lessons in physiotherapy education

Providing high-quality feedback is essential for developing clinical skills in health science students, but it remains a challenge for many instructors. Artificial intelligence (AI) has proven to be a transformative tool for supporting feedback processes in higher education. However, more empirical evidence is needed to demonstrate its impact on health sciences education. This study evaluated the use of an AI-based tool integrated into a locally developed feedback-oriented platform and used in a course involving 96 physiotherapy students and nine instructors in a long-established Chilean university. In phase one, students uploaded a self-recorded video of a simulated clinical interview and received instructor feedback without AI support. In phase two, students submitted a new video, and instructors used the AI tool to review and refine their comments before sharing them. A total of 1435 feedback comments were analysed across five evidence-based quality criteria. Results showed significant improvements (p < 0.01) in criteria: direct observation, specific information, positive reinforcement, and action planning. The other criterion, providing suggestions for improvement, was more influenced by students’ performance. Overall, the AI tool helped instructors improve the quality of their feedback without replacing their role, demonstrating its potential to strengthen clinical education by supporting educators in higher education contexts.