Trust in AI-generated medical videos and professional identity among medical students: the mediating role of emotional resonance

Sažetak

Objective To examine the association between trust in medical short videos produced using generative artificial intelligence (AI) and professional identity among medical students, assess the mediating role of emotional resonance and moderation by academic year and self-reported AI recognition ability. Methods A cross-sectional questionnaire survey included 1,024 medical students at one medical university. Mediation and moderated mediation were examined using PROCESS Models 4 and 7. Results Trust was positively correlated with emotional resonance ( r = 0.234, P < 0.001) but not with professional identity ( r = 0.035, P = 0.269). Emotional resonance was positively correlated with professional identity ( r = 0.312, P < 0.001). After adjustment for covariates, the indirect effect through emotional resonance was small ( B = 0.040, 95% CI [0.023, 0.061]; completely standardized estimate = 0.063, 95% CI [0.036, 0.094]), whereas the direct effect was not significant ( B = −0.022, 95% CI [−0.061, 0.017], P = 0.264). The interaction between academic year and trust was significant ( B = −0.135, 95% CI [−0.232, −0.039], P = 0.006). The index of moderated mediation was −0.037 (95% CI [−0.074, −0.003]), with a larger indirect effect among lower-year students (0.065) than higher-year students (0.028). The interaction between self-reported AI recognition ability and trust was not significant ( B = 0.005, 95% CI [−0.036, 0.045], P = 0.828). Conclusion Trust in generative AI medical short videos showed a small indirect association with professional identity through emotional resonance. This indirect association was stronger among lower-year students. Given the cross-sectional design, measurement limitations, and possible common method bias, these findings warrant cautious interpretation and do not establish a causal mechanism.