Exploring the potential for generative AI to facilitate medical students’ use of evidence-based learning strategies

Abstract

Objective Effective learning strategies, such as retrieval practice, spacing and interleaved practice, support meaningful knowledge construction and long-term retention. However, students often do not use these strategies because of misconceptions, perceived time and effort. This study aimed to explore whether and how an AI review module for a pathology course, integrated in a learning management system, influences third-year preclinical students’ learning strategies at Tsinghua University. Design This study used a sequential explanatory mixed-methods design. For the quantitative strand, pre-course and post-course questionnaires assessed students’ self-reported learning strategies and voluntary use of the AI review module. Based on the usage log data, students were categorized into high- and low- usage groups for quantitative and qualitative analysis. Non-parametric statistical analyses were conducted when appropriate. For the qualitative strand, interviewees were selected using maximum variation sampling based on their AI review module usage. Semi-structured one-on-one interviews were conducted and qualitative descriptive approach were performed. Results Fifty students completed the pre-course questionnaire and took the final exam. The most frequently used learning strategies were non-evidence-based approaches. Thirty-seven (74%) students used the AI review module and collectively answered 1804 questions. Two-sample Wilcoxon rank-sum tests revealed no statistically significant associations between AI module usage and changes in learning strategies or pathology final exam scores. In the interviews, a few students recognized and intentionally applied the evidence-based learning strategies embedded in the AI module, whereas some other students still used this AI tool for rote memorization and cramming before exams, which was inconsistent with evidence-based learning strategies. Conclusion The AI-based review module, although designed according to evidence-based learning strategies, did not increase the use of evidence-based strategies or improve exam scores at the class level. Our findings suggest that, as AI tools become more widespread in medical education, explicit attention to learning sciences and learning strategies is essential, because students are likely to use AI according to their pre-existing study habits, and AI may magnify the effects, either beneficial or detrimental, of those habits.