Beyond acceptance: bounded use of AI and VR in orthopedic education

The educational roles of artificial intelligence (AI) and virtual reality (VR) in orthopedic training remain insufficiently defined. Rather than examining acceptance alone, we investigated how students delineate trust boundaries, role allocation, and implementation priorities for bounded, faculty-supervised use of AI/VR in orthopedic education.

We conducted an exploratory cross-sectional online survey among Years 4-8 students in the eight-year program of clinical medicine at Peking Union Medical College using a study-specific questionnaire informed by AI- and simulation-enhanced medical education. Approximately 300 students were invited; 56 valid responses were analyzed (response proportion, 18.7%). The questionnaire assessed prior exposure to AI/VR, acceptance, perceived value, trust-related attitudes, application preferences, substitution boundaries, and risk perceptions.

Overall, 83.9% of respondents supported or strongly supported further AI/VR integration in orthopedic teaching. Acceptance showed the highest composite score (4.25 ± 0.73), followed by perceived value (4.22 ± 0.61), educational trust (4.02 ± 0.49), and risk perception (3.80 ± 0.53). Support for AI-assisted tools and VR-based simulation was similarly high (4.29 ± 0.73 vs. 4.32 ± 0.83, p = 0.694). The strongest endorsements were for continued in-person teacher guidance (4.59 ± 0.53) and teacher review of AI/VR content before formal use (4.52 ± 0.57), whereas trust in AI-generated recommendations for real patient decision-making was the lowest-rated item (3.11 ± 1.11). Students consistently favored augmentation over replacement: AI was preferred for knowledge summarization, reasoning support, and examination preparation, whereas VR was preferred for anatomy learning, fracture classification, surgical pathway visualization, and pre-training. Prior VR exposure was associated with greater implementation support (4.41 ± 0.51 vs. 3.85 ± 0.90, p = 0.014).

Students supported AI/VR in orthopedic education, but within clear pedagogical and governance boundaries. Their responses suggest a preliminary, role-specific implementation model in which AI primarily enhances cognitive learning and VR primarily supports spatial-procedural training, under sustained faculty oversight. These perception-based findings should be tested in longitudinal or interventional curriculum studies.