Medical schools are extraordinarily good at teaching future physicians what they need to know. The harder question is whether they prepare them for the moment when knowledge no longer provides a single answer.
Medical education has transformed an enormous and constantly expanding body of scientific knowledge into something that can be taught, examined, and applied to patients. Students learn anatomy and physiology, pathology and pharmacology, diagnosis and treatment. They learn to recognize patterns, interpret findings, and distinguish the probable from the improbable. By graduation, they know vastly more medicine than physicians could have imagined only a few generations ago. And yet many encounter the same surprise when they enter clinical practice. The difficulty of medicine is not simply that there is more to know. It is that the nature of the problem changes.
Examinations necessarily reward correct answers. Clinical reality frequently offers several imperfect ones. Patients arrive with incomplete information, contradictory findings, multiple diseases, personal preferences, and circumstances never represented in the textbook. A guideline may describe what should happen in most cases without determining what should happen in this one. An elderly patient with heart failure, renal impairment, and recurrent falls does not present as a multiple-choice question. Whether another diagnostic procedure will actually help cannot be answered by medical evidence alone. The decision is medical, but also human, ethical, and increasingly systemic. This is where knowledge becomes judgment.
Judgment is sometimes treated as a synonym for experience, as though it simply accumulates with years of practice. But it is more demanding. It means deciding which information matters, balancing risks that cannot be eliminated, recognizing when the technically possible is not necessarily clinically appropriate, and accepting responsibility when no option is unquestionably correct. Much of what distinguishes an experienced physician is still acquired through the hidden curriculum: observing how a senior physician handles uncertainty, listening to how difficult choices are explained to patients, and understanding why someone departs from a familiar pathway.
Real physician voices, twice a week
Free, and one click to unsubscribe.
The problem with a hidden curriculum is that what it teaches depends too much on where and with whom a young physician happens to train. It also teaches lessons we may not intend. Young doctors quickly discover that certainty sounds more authoritative than uncertainty. Efficiency is visible; reflection often is not. Following an established pathway is easier to defend than departing from it. Saying “I don’t know” can feel like an admission of incompetence even when it is the most intellectually honest starting point. Uncertainty can therefore become something to escape rather than something to manage.
Physicians increasingly work in systems in which clinical decisions depend on teams, workflows, resources, and technologies. A brilliant individual decision can still produce a poor outcome if the surrounding system fails. Yet medical education remains much better at preparing doctors for individual clinical performance than for responsibility within complex systems.
AI makes this gap particularly visible. Much of the discussion about AI in medical education begins with the knowledge question: What must students still memorize if information can be retrieved instantly? Which diagnostic tasks will algorithms perform? How should examinations change when generative systems can produce sophisticated answers within seconds?
AI will become increasingly good at activities around which much of formal education has traditionally been organized: retrieving information, identifying patterns, comparing possibilities, summarizing evidence, and generating plausible recommendations. That does not make medical knowledge less important. A physician who cannot understand the medicine cannot reliably recognize when a machine is wrong. But it changes where the educational challenge lies.
The central question may no longer be simply how much physicians need to know in an age of intelligent machines. It may be how they learn to decide when knowledge (whether generated by a textbook, guideline, colleague, or algorithm) does not settle the matter. AI systems will often provide probabilities rather than certainty and may identify patterns whose reasoning is difficult to reconstruct. They will sometimes disagree with the clinician or the patient’s expectations. Someone will still have to determine whether a recommendation fits the person rather than merely the data, explain the decision, and ultimately accept responsibility for it.
The future of medical education should therefore not be built by adding another layer of content to an already overcrowded curriculum. Medical schools need to become more deliberate about capabilities too often expected to emerge by themselves.
Students should encounter cases in which several answers are reasonable and be asked not only what they would do, but why. Clinical teachers should make their reasoning visible, including the uncertainty they considered and alternatives they rejected. Young physicians should learn that acknowledging uncertainty is not the opposite of competence but often a prerequisite for responsible judgment. Medical education should also expose students earlier to the systems within which their decisions will have consequences.
None of this requires abandoning the scientific foundations of medical education. The opposite is true. In an era of misinformation, information abundance, and increasingly persuasive AI-generated content, deep scientific understanding becomes even more important. But science and judgment perform different functions. Science tells us what we know. Judgment becomes necessary when what we know is no longer sufficient to tell us what to do. Medical schools cannot reproduce every situation their graduates will encounter. Nor should they try. But we can stop treating the most difficult parts of becoming a physician as lessons that somehow teach themselves.
The future physician will practice with access to more knowledge, data, and computational intelligence than any generation before. The defining educational challenge will not be ensuring that doctors can compete with machines in producing answers. It will be preparing them for the moments when an answer is only the beginning of a decision. Medicine will remain a science. But the practice of medicine will remain an act of responsibility. And responsibility begins precisely where the correct answer ends.
Mathias Goyen is a radiology professor.
