This is the fourth piece in our series on strategic planning in health care. Having looked at systemic complexity, blurring boundaries, and the massive implications of AI, we turn to an area where all these forces collide: medical education. Medical schools, residency and fellowship programs must educate future physicians now for careers extending far into a future in which we cannot confidently describe the health care system, the technology physicians will use, the places where care will be delivered—or even exactly what physicians will be expected to do. This post considers the challenge of preparing students, residents, and fellows for a future system we cannot yet definitively describe.
Educating for a profession that is fundamentally changing
Throughout this piece we draw on FSG’s 14 years of strategic work with medical education organizations, and although we suggest a sharp contrast between what has been and what could be, we should add here that many of the issues and ideas in this essay have been part of this work and are already being addressed. Change in systems as complex as health care and medical education is evolutionary rather than revolutionary and it will not follow a straight path.
For most of modern medicine’s history, medical education has delivered an enormous body of scientific and clinical knowledge and experience to students, residents, and fellows, taught them how to apply it, and progressively exposed them to patients until they were prepared to practice independently. That framework isn’t disappearing, but almost every part of it is being challenged.
AI can already retrieve and synthesize medical information almost instantaneously. Diagnostic systems can recognize patterns in images and data. Patients arrive armed with information—and increasingly AI-generated interpretations of their symptoms. Care is migrating beyond hospitals and physicians’ offices into homes, retail settings, virtual platforms and community organizations. At the same time, the boundaries among physicians and other health professionals continue to blur.
From knowing to judging?
It would be wrong to assume physicians will simply need to know less. A physician cannot intelligently challenge an AI recommendation without sufficient knowledge to recognize that something may be wrong. But ubiquitous access to machine intelligence could change the relative importance of different capabilities.
Medical education has necessarily devoted enormous resources to acquiring, retaining and demonstrating knowledge, but in an AI-enabled environment, the more valuable skill may become knowing how to interrogate information, recognize uncertainty, integrate conflicting evidence, exercise judgment, and make decisions in a patient-centric manner.
That could additionally elevate capabilities sometimes treated as the softer edges of medical education: communication, ethics, teamwork, adaptability, critical thinking and professional judgment.
The paradox of AI may be that the more technologically sophisticated medicine becomes, the more valuable distinctly human capabilities become.
Who does what?
A further complication is that physicians are not evolving independently of the rest of the health care workforce.
Advanced practice nurses, physician associates, pharmacists, psychologists, community health workers and other professionals are already assuming responsibilities once associated primarily with physicians. AI potentially accelerates the process, which creates questions extending well beyond curriculum design:
- Will professional boundaries continue to blur?
- What competencies genuinely require a physician?
- Will education become substantially more interprofessional?
- Will AI allow generalists to manage greater complexity, or encourage still greater specialization?
- And could some medical roles eventually require less training while others require more?
Different answers lead to very different medical education systems.
Training where medicine happens
The setting for medical education presents another challenge, as much of clinical training remains centered on hospitals and academic medical centers. But an increasing share of medicine is being delivered elsewhere. If the future of care is more ambulatory, virtual, home-based, preventive and community-oriented, medical education will presumably have to follow it.
Technology could simultaneously transform clinical training itself. Sophisticated simulation and AI-generated patients may eventually allow students to encounter hundreds or thousands of clinical situations—including rare diseases and unusual presentations—that conventional clinical rotations could never reliably provide. [This is especially true today for trainees in non-urban settings, with limited exposure to diverse patient populations.]
This raises the possibility of future medical education being simultaneously more technological and more human, using machines extensively to teach the science and practice of medicine while preserving (and perhaps enhancing) encounters with real patients for precisely those things machines cannot reproduce.
How many physicians will we need?
Which brings us to the ultimate strategic uncertainty: supply and demand. The United States faces a rapidly aging population, geographic maldistribution of physicians and persistent shortages in some specialties and communities. Medical student debt reinforces some of these shortages, by making higher-paying specialties more attractive than primary care.
So, will more physicians need to be educated and trained – or will AI dramatically increase physician productivity? Preventive care and remote monitoring could also reduce some demand for acute medical intervention. Scope-of-practice boundaries could continue to blur, with more routine care delivered by other professionals, supported by increasingly capable technology. However this may take a while to find a ‘sweet spot’ as less physician early access could result in higher downstream costs due to delayed diagnoses, potentially excessive testing, and loss of continuity in relationships
The question therefore might not be how to produce substantially more physicians, but rather what kinds of physicians are needed, where they are needed and what activities most require their expertise.
We have touched on some profoundly different futures – each implying somewhat different physician characteristics and therefore different medical education models.
However our approach does not require planners to choose among them, but instead encourages them to explore a more useful question:
Are there strategies that would make sense across very different futures?
In the work that FSG has done in health care we have explored a variety of scenario themes — including for example radical privatization of health care on one extreme and single payor government system on the other. The way some of the current trends play out will depend on which way the system as a whole evolves over the next couple of decades.
But some answers may prove surprisingly robust.
The education system will need to prepare for and adapt to the shifting career needs of physicians, so that mid-career specialty changes are both possible and not overly burdensome.
Physicians will probably need to become comfortable working alongside intelligent machines without becoming dependent upon them. They will need to understand systems as well as individual patients. They will need to collaborate across increasingly porous professional boundaries. They will need to continue to make decisions under uncertainty.
And they will need to preserve the judgment, accountability, ethics and human connection upon which medical professionalism ultimately depends.
Future medical schools and residency programs, therefore, may not simply be today’s institution with more AI in the curriculum but could eventually represent a more fundamental rethinking of what society needs a physician to be.
We don’t yet know the answer.
And that is precisely why medical education needs scenario thinking.