JMIR Publications

ABSTRACT

Background:

Generative artificial intelligence (GAI) is increasingly used in medical education, but concerns remain about inaccurate information, privacy, academic integrity, and overreliance. How medical students balance the perceived benefits and risks of GAI remains insufficiently understood.

Objective:

This study examined medical students’ behavioral intention to use GAI using an extended Unified Theory of Acceptance and Use of Technology framework incorporating perceived risk and technology anxiety, and explored risk-aware adoption through qualitative interviews.

Methods:

Explanatory sequential mixed-methods design was used. In the quantitative phase, 572 valid survey responses from medical students in China were analyzed. Measures included performance expectancy, effort expectancy, social influence, facilitating conditions, perceived risk, technology anxiety, and behavioral intention. Reliability, confirmatory factor analysis, regression, mediation, moderation, and subgroup analyses were conducted. In the qualitative phase, semi-structured interviews with 15 purposively selected respondents were analyzed thematically.

Results:

Among the 572 respondents, 356 (62.2%) reported using GAI at least weekly. Social influence was the strongest positive predictor, followed by performance expectancy, effort expectancy, and facilitating conditions. Perceived risk and technology anxiety were not significant independent predictors, and technology anxiety did not significantly mediate or moderate the perceived risk–intention relationship. Daily users reported both the highest perceived risk and the highest behavioral intention, suggesting a risk-aware rather than risk-ignorant adoption pattern. Qualitative analysis identified five themes: efficiency-driven adoption, concrete risk awareness, verification and boundary-setting, cautious rather than avoidant responses to anxiety, and the need for institutional guidance and AI literacy training.

Conclusions:

Medical students’ intention to use GAI was mainly associated with acceptance-related factors rather than risk-related barriers. Rather than a simple benefit–risk trade-off, the findings reveal a pattern of risk-aware adoption: students sustained high intention to use GAI not because they ignored risks, but because they actively managed them through verification, boundary-setting, and selective trust. This reframes the role of perceived risk from a barrier to adoption to a driver of responsible use. Clinical Trial: null