Chatgpt-Assisted Learning Versus Traditional Learning Methods in Medical Education: A Systematic Review and Meta-Analysis
Keywords:
ChatGPT; generative artificial intelligence; medical education; traditional teaching; knowledge acquisition; knowledge retention; clinical skills; academic performance; randomized controlled trials; systematic review; meta-analysis.Abstract
Background: The integration of ChatGPT and generative artificial intelligence (AI) into medical education has introduced new opportunities for personalized learning, clinical reasoning, formative assessment, and knowledge acquisition. Although AI-assisted instruction is increasingly adopted, its comparative effectiveness against traditional teaching approaches remains uncertain, particularly regarding knowledge retention, clinical competence, and academic performance.
Objective: To systematically evaluate the effectiveness of ChatGPT-assisted learning compared with traditional teaching methods among undergraduate and postgraduate medical trainees, focusing on knowledge acquisition, clinical skills, academic performance, knowledge retention, learner satisfaction, and educational outcomes.
Methods: A systematic review of comparative educational studies was conducted using literature identified through PubMed/MEDLINE, Scopus, Embase, Web of Science, and supplementary reference searches through September 2026. Randomized controlled trials and comparative studies evaluating ChatGPT-assisted instruction against conventional teaching methods were considered. Study characteristics, intervention strategies, educational outcomes, and methodological limitations were assessed. Published meta-analytic effect estimates were synthesized narratively, with heterogeneity evaluated using the I² statistic.
Results: Database searches identified 1,386 records, while reference lists and publisher sources contributed 29 additional records, yielding 1,415 records. Following removal of 254 duplicates, 1,161 records underwent title and abstract screening, of which 1,073 were excluded. Eighty-eight full-text articles were assessed for eligibility, and 80 were excluded because of unsuitable comparisons (24), unavailable learner outcomes (19), ineligible study designs (15), nonmedical populations (12), or overlapping reports (10). Eight studies were included in the qualitative synthesis. Published meta-analyses demonstrated variable educational benefits. A meta-analysis of 11 randomized controlled trials involving 786 students reported no statistically significant improvement in knowledge acquisition (standardized mean difference [SMD] 0.27; 95% confidence interval [CI] −0.31 to 0.85), although practical skills improved (SMD 0.63; 95% CI 0.10–1.16). Another synthesis of 14 randomized trials involving 1,116 learners demonstrated improvements in knowledge acquisition (SMD 0.36; 95% CI 0.24–0.49) and clinical skills (SMD 0.78; 95% CI 0.57–0.99). A larger meta-analysis of 20 randomized trials involving 1,413 learners reported favorable effects on knowledge acquisition (SMD 0.99; 95% CI 0.49–1.49) and clinical skills (SMD 1.09; 95% CI 0.81–1.38), with substantial heterogeneity in knowledge outcomes (I² = 86.7%). Individual trials demonstrated inconsistent improvements in delayed knowledge retention and independently assessed bedside clinical performance. The available trial-level data were insufficient for an independent assessment of funnel-plot asymmetry or publication bias.
Conclusion: ChatGPT-assisted learning may improve knowledge acquisition, learner engagement, and selected clinical skills compared with traditional instructional approaches. However, differences in intervention design, assessment methods, follow-up duration, and substantial between-study heterogeneity limit the certainty and generalizability of these findings. Evidence supporting sustained knowledge retention and improved real-patient clinical performance remains inconclusive. ChatGPT should be considered a supplementary educational tool integrated with faculty supervision, structured feedback, and independent competency assessment rather than a replacement for conventional medical teaching.

