JMIR Publications

ABSTRACT

Background:

Generative artificial intelligence (GenAI) has been rapidly adopted by medical students, yet little is known about whether institutional policies and guidance are visible to students or how students judge acceptable use.

Objective:

This study examined medical students’ GenAI-use patterns, their awareness of and perceptions toward institutional AI-use policy and guidance, and how these constructs are associated with one another and with prior formal AI or machine-learning (AI/ML) coursework.

Methods:

We conducted a cross-sectional survey of 298 medical students using an anonymous questionnaire assessing GenAI-use patterns and 4 conceptually defined dimensions: policy awareness, acceptability of GenAI use, institutional support and guidance, and stakeholder responsibility. Associations between use patterns and policy-related dimensions were examined using Spearman correlations, and group differences by prior AI/ML coursework were examined using Mann–Whitney U tests.

Results:

GenAI use was widespread and frequent despite limited formal AI/ML training: 95% of students used GenAI at least weekly, and 71% had no prior AI/ML coursework. Only one-third of students believed their university had a formal AI-use policy, fewer than half had encountered an official statement, and awareness of specific AI-related institutional resources was low. Students perceived supportive uses of GenAI as more acceptable than uses involving limited disclosure, transparency, or original contribution. A pronounced gap separated strong endorsement of institutional preparation for responsible AI use from low perceived access to guidance. Responsibility was viewed as shared, with medical schools rated most responsible. More frequent GenAI use was associated with lower policy awareness and weaker perceived institutional support and guidance, whereas prior AI/ML coursework was associated with higher scores across all policy-related dimensions.

Conclusions:

These findings reveal a practice–policy gap: everyday GenAI use has outpaced the visibility and uptake of institutional guidance, and the students who use GenAI most are the least aware of existing guidance. Medical schools should provide clear, accessible, course-embedded guidance introduced early, revisited throughout training, and shared across stakeholders.