ARTIFICIAL INTELLIGENCE (AI) TOOL USE VERSUS EXPERT-LED INSTRUCTION IN PRECLINICAL MEDICAL EDUCATION: A CROSSSECTIONAL SURVEY OF STUDENT EXPERIENCE, PERCEPTIONS, AND ACADEMIC PERFORMANCE
TL;DR
These findings support incorporating AI literacy and ethics into preclinical curricula while preserving the strengths of expert-led instruction.
Abstract
Background: Artificial Intelligence (AI) is increasingly integrated into preclinical medical education. Despite rapid adoption, comparative data on student satisfaction, perceptions, learning preferences, and academic outcomes remain limited. This study aimed to evaluate these outcomes among preclinical medical students. Methods: A cross-sectional study was conducted among preclinical medical students at Phramongkutklao College of Medicine, Bangkok, Thailand (June 2024–May 2025). A validated questionnaire was administered. Academic performance was measured by cumulative grade point average (GPAX) and stratified by AI usage intensity: more-AI-using (>5 h/week) and less-AI-using (≤5 h/week). Statistical analyses included the Wilcoxon signed-rank test and an independent t-test (p < 0.05). Results: A total of 220 students completed the questionnaire (Year 2: 44.1%; male: 55.9%; mean age: 19.4 ± 1.6 years). 90.9% had previously used AI tools in preclinical studies, most commonly for 3–5 hours per week. Expert-led instruction was rated significantly higher than AI for effectiveness, preclinical confidence, clinical readiness, and overall satisfaction. 90.5% preferred expert-led instruction as the primary modality; however, 87.7% endorsed an integrated AI–expert framework (concurrent or AI after expert). Despite 93.6% of students reporting that AI positively impacted their learning outcomes and 82.7% reporting improved problem-solving skills, GPAX did not differ significantly between the more AI-using (>5 hours/week, 3.21 ± 0.64) and less AI-using (≤5 hours/week, 3.27 ± 0.56) groups (p = 0.100). Conclusion: Preclinical medical students used AI tools extensively but rated expert-led instruction higher in terms of overall satisfaction. No significant association was found between self-reported AI-use intensity and academic performance. Students favored a blended model that positions AI as a complement to expert-led instruction. These findings support incorporating AI literacy and ethics into preclinical curricula while preserving the strengths of expert-led instruction.