The rapid emergence of generative artificial intelligence models, such as ChatGPT and successors like GPT-4, has triggered a seismic shift in medical education and practice. In a special podcast series revisiting the book ´The AI Revolution in Medicine,´ Microsoft Research president Peter Lee interviews Dr. Morgan Cheatham and Daniel Chen—two rising physician-leaders with deep interests in both clinical practice and technology. Their experiences highlight both the transformative potential and the unique challenges brought about by artificial intelligence tools in how future doctors are trained and how they care for patients.
Cheatham, now a resident at Boston Children’s Hospital and a partner at Breyer Capital, reflects on his journey intersecting computer science, venture capital, and clinical medicine. He describes how generative artificial intelligence has become a powerful adjunct for studying, simulating patient encounters, and supporting clinical decision-making. Cheatham participated in early research proving tools like ChatGPT could pass the United States Medical Licensing Examination, a breakthrough that provoked serious reconsideration of traditional curricula and assessment methods. As students increasingly rely on such models for explanations, second opinions, and efficiency at the bedside, Cheatham stresses the urgent need to update medical education to incorporate responsible artificial intelligence use. He advocates for formal training on the evaluation, strengths, and limitations of these tools, and argues that both old and new skills—like crafting clinical notes unaided and then hybridizing with artificial intelligence—should have a place, emphasizing adaptability and critical assessment.
Daniel Chen, a second-year medical student at the Kaiser Permanente School of Medicine, describes generative artificial intelligence as an omnipresent study aid, research assistant, and clinical tutor. He underscores how widespread use among students is reshaping learning styles, with artificial intelligence acting as an explainer, abbreviation translator, and brainstorming partner for differentials or lab orders. However, Chen and his peers remain alert to pitfalls, including the risk of overreliance and erosion of critical reasoning, as well as the real challenges of assessing trustworthiness amid sources of hallucination. Despite the technology’s surge, formal curriculum change lags, with most artificial intelligence-related innovation being student-led or ad hoc. Both Cheatham and Chen agree that while the accrediting bodies and faculty are cautious, a grassroots movement of students is likely to drive future standards and best practices for integrating artificial intelligence safely and effectively into physician training.
The discussion also extends to broader cultural and ethical questions: how will responsibilities shift as artificial intelligence moves from back-office to direct patient care applications, and as patients themselves come armed with their own machine-generated medical hypotheses? Both interviewees believe that the ability to prompt, interpret, and collaborate with artificial intelligence systems will soon be foundational to clinical competence, and that the evolving role of the physician must not only accommodate technology but actively shape it to keep compassion and judgment at the center of care. Ultimately, the episode paints an optimistic vision: with critical training and prudent oversight, generative artificial intelligence can catalyze a more insightful, patient-responsive, and continuously learning health system.
