Applications, Challenges, and Prospects of Generative Artificial Intelligence Empowering Medical Education: Scoping Review
Bibliographic record
Abstract
Background: Nowadays, generative artificial intelligence (GAI) drives medical education toward enhanced intelligence, personalization, and interactivity. With its vast generative abilities and diverse applications, GAI redefines how educational resources are accessed, teaching methods are implemented, and assessments are conducted. Objective: This study aimed to review the current applications of GAI in medical education; analyze its opportunities and challenges; identify its strengths and potential issues in educational methods, assessments, and resources; and capture GAI's rapid evolution and multidimensional applications in medical education, thereby providing a theoretical foundation for future practice. Methods: This scoping review used PubMed, Web of Science, and Scopus to analyze literature from January 2023 to October 2024, focusing on GAI applications in medical education. Following PRISMA-ScR (Preferred Reporting Items for Systematic Reviews and Meta-Analyses Extension for Scoping Reviews) guidelines, 5991 articles were retrieved, with 1304 duplicates removed. The 2-stage screening (title or abstract and full-text review) excluded 4564 articles and a supplementary search included 8 articles, yielding 131 studies for final synthesis. We included (1) studies addressing GAI's applications, challenges, or future directions in medical education, (2) empirical research, systematic reviews, and meta-analyses, and (3) English-language articles. We excluded commentaries, editorials, viewpoints, perspectives, short reports, or communications with low levels of evidence, non-GAI technologies, and studies centered on other fields of medical education (eg, nursing). We integrated quantitative analysis of publication trends and Human Development Index (HDI) with thematic analysis of applications, technical limitations, and ethical implications. Results: Analysis of 131 articles revealed that 74.0% (n=97) originated from countries or regions with very high HDI, with the United States contributing the most (n=33); 14.5% (n=19) were from high HDI countries, 5.3% (n=7) from medium HDI countries, and 2.2% (n=3) from low HDI countries, with 3.8% (n=5) involving cross-HDI collaborations. ChatGPT was the most studied GAI model (n=119), followed by Gemini (n=22), Copilot (n=11), Claude (n=6), and LLaMA (n=4). Thematic analysis indicated that GAI applications in medical education mainly embody the diversification of educational methods, scientific evaluation of educational assessments, and dynamic optimization of educational resources. However, it also highlighted current limitations and potential future challenges, including insufficient scene adaptability, data quality and information bias, overreliance, and ethical controversies. Conclusions: GAI application in medical education exhibits significant regional disparities in development, and model research statistics reflect researchers' certain usage preferences. GAI holds potential for empowering medical education, but widespread adoption requires overcoming complex technical and ethical challenges. Grounded in symbiotic agency theory, we advocate establishing the resource-method-assessment tripartite model, developing specialized models and constructing an integrated system of general large language models incorporating specialized ones, promoting resource sharing, refining ethical governance, and building an educational ecosystem fostering human-machine symbiosis, enabling deep tech-humanism integration and advancing medical education toward greater efficiency and human-centeredness.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame distilled prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. Learned from the 10,348 direct Codex labels and 10,348 direct Gemma labels. Candidate is the union of thresholded teacher heads; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels or direct frontier model labels.
Codex and Gemma teacher scores by category
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.001 | 0.003 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.001 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one teacher head, not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".