Artificial Intelligence Use and Perceptions Among Three Medical Student Cohorts: Cross-Sectional Comparative Survey Study

Artificial intelligence (AI) tools, particularly large language models, are increasingly used by medical students, yet patterns of use, perceived reliability, verification practices, and educational expectations may vary across educational contexts. We conducted an exploratory, cross-sectional, anonymous web-based survey at the Albert Szent-Györgyi Medical School, University of Szeged, Hungary, comparing first-year students in the Hungarian-language program (MedHU), first-year students in the English-language program (MedEN), and second- to fifth-year Hungarian-language students enrolled in an AI-focused elective course (MedAI). The author-developed 20-item questionnaire assessed AI use, familiarity, perceived reliability, verification practices, limitations, and educational expectations. Of 550 eligible students, 340 submitted questionnaires and 301 met predefined data-quality criteria. Academic AI use was reported by 81.9% of MedHU, 85.6% of MedEN, and 87.5% of MedAI participants. Use-frequency distributions differed among cohorts (χ2(6) = 64.63, p < 0.001, φ = 0.463); daily use was reported by 43.3% of MedEN, 7.8% of MedHU, and 12.5% of MedAI participants, while familiarity with AI-related terminology varied by term across cohorts. Across cohorts, students generally regarded AI-generated medical information as requiring verification and viewed AI as supplementary to conventional educational resources. These findings identify priorities for AI literacy, particularly critical appraisal, source verification, and responsible use.