Artificial Intelligence in Medical Education: A Narrative Review of Current Applications and Future Perspectives

Abstract

Introduction and purpose. Artificial intelligence has gained growing attention in medical education, especially since ChatGPT appeared in late 2022. Large language models proved capable of reaching or exceeding passing thresholds on USMLE-style questions or official practice materials, which sparked both enthusiasm and concern among educators. This review aims to present how AI tools are currently used across the medical education continuum, what problems they create and where the field may be heading.

Brief description of the state of knowledge. Machine learning and natural language processing are already being used in preclinical teaching, especially in anatomy and pharmacology, through adaptive quizzes, intelligent tutoring, and question generation. In clinical training, AI-powered virtual patients allow students to practise history-taking and diagnostic reasoning in a safe setting, while AI-enhanced surgical simulators give objective, real-time feedback on technical skills. Large language models are also being tested as item writers for medical examinations, with early studies showing acceptable question quality after expert review. At the same time, serious problems remain: AI systems can produce confident but wrong medical information, student overreliance may weaken independent reasoning, academic integrity is harder to enforce and algorithmic bias may disadvantage certain groups.

Summary. AI offers real opportunities to personalise learning, broaden access, and improve assessment in medical education. Realising these benefits without harm will require clear institutional policies, faculty development, rigorous outcome research, and a commitment to preserving the human side of medicine.