AI Is Changing What It Means to Be a Doctor: Are We Ready?
Artificial intelligence is no longer waiting on the horizon of medicine. It has already entered the clinic, the emergency department, and the electronic health record. Systems now diagnose complex cases more accurately than many physicians, generate detailed treatment plans, take patient histories with greater consistency, and produce clinical notes while the doctor is still speaking with the patient. The classical image of the doctor as the sole repository of medical knowledge is fading. The urgent question is whether the profession, medical schools, and health systems are prepared for what comes next.
In a TEDx talk at NYU Langone Health, physician and educator Marc Triola captured the moment with unusual clarity. Medical AI systems, he noted, are already smarter than human doctors in several core tasks. Research shows they often outperform physicians at diagnosing patients, formulating treatment plans, and even conducting the initial history. In some states, AI tools can renew prescriptions without direct human oversight. At the same time, medical knowledge continues to expand at a pace no human can match—thousands of new studies appear every day. The traditional model of medical education, built around memorizing vast quantities of information, no longer fits the realities of modern practice.
The Evidence Is No Longer Theoretical
The performance gap is measurable. Google’s Articulate Medical Intelligence Explorer has demonstrated superior ability to elicit patient complaints, review symptoms, and manage conditions over multiple visits compared with primary care physicians. Advanced language models have correctly identified final diagnoses in complex real-world cases at rates several times higher than groups of internal medicine doctors. Imaging AI now matches specialist accuracy on three-dimensional scans across MRI, CT, ultrasound, and retinal imaging, often in a fraction of the time. Ambient documentation tools listen to conversations and produce complete notes, suggested codes, and follow-up plans, cutting the hours physicians once spent typing after clinic.
These tools are already in widespread use. Surveys in 2026 show that the large majority of physicians now use AI in their professional work, more than double the rate of just a few years earlier. The most immediate gains have come in administrative relief. Doctors have long spent roughly two hours on paperwork for every hour of direct patient care. AI scribes and inbox triage systems are reclaiming significant portions of that time. Many clinicians report more eye contact with patients, reduced burnout symptoms, and the capacity to see additional people each week.
Yet the most unsettling finding may be about empathy. In blinded evaluations across oncology, neurology, mental health, and general medicine, written responses generated by chatbots have frequently been judged more empathetic than those written by physicians. This does not mean machines feel compassion. It reveals how far clinical practice has drifted under the weight of documentation demands, productivity metrics, and fragmented workflows. When the human doctor is forced to spend more time looking at a screen than at the person in front of them, a well-trained language model can appear more present.
A Profession in Transition
AI is not simply adding new instruments to the doctor’s bag. It is reshaping the job itself. The physician reduced to a clerk—processing prior authorizations, coding visits, answering endless inbox messages—is already partially replaceable, because those tasks are precisely what current systems handle best. Freeing clinicians from that burden creates the possibility of restoring the parts of medicine that remain distinctly human: navigating uncertainty, making ethical judgments, building trust, and exercising the kind of clinical wisdom that comes from years of experience with real patients.
This shift, however, carries real dangers. One is “AI-induced deskilling.” As systems become more reliable, the incentive to maintain independent diagnostic and reasoning skills weakens. Some researchers now argue that hybrid human-AI teams can underperform pure AI systems in certain diagnostic settings, and that autonomous models may outperform both physicians and human-AI combinations in key tasks by 2030. Others warn that the tools are being designed and deployed largely without practicing physicians at the center, further eroding the limited professional autonomy doctors still retain. When algorithms generate notes, suggest orders, and flag risks before the physician has finished examining the patient, the space for independent clinical judgment narrows.
Medical education faces an equally serious challenge. Traditional curricula emphasize knowledge acquisition and pattern recognition. The emerging role demands something different: the ability to critically evaluate AI recommendations, recognize when models fail or introduce bias, maintain practical skills as a safety net, and lead the responsible integration of these systems. A few institutions, including NYU Grossman School of Medicine, are already redesigning training around precision medical education that uses AI to personalize learning pathways. Broader reform across the profession remains uneven and slow.
The Readiness Test
The global context makes delay costly. Health systems face projected shortfalls of millions of workers by 2030. Burnout remains stubbornly high. AI is increasingly framed not as a replacement strategy but as a retention strategy—one that can preserve careers by reducing the clerical load that drives people out of practice. High-impact applications such as ambient documentation, coding support, demand prediction, and claims processing can return time to the bedside, teaching, and leadership. Poorly designed implementation, however, can simply shift burdens, erode confidence, and widen existing gaps in quality and access.
True readiness requires more than buying software. It demands clear rules for liability when AI systems err, transparency about training data and validation, safeguards against bias that could worsen health disparities, and genuine physician involvement in how tools are selected and used. Patients must understand when AI is participating in their care and retain meaningful avenues for human oversight. Technology companies should not unilaterally determine how care is delivered or how trust is maintained.
Two futures are visible. In one, AI absorbs the cognitive and administrative overload, physicians reclaim time for presence and judgment, and medicine becomes both more precise and more human. In the other, the profession adapts too slowly, clinical skills atrophy, autonomy continues to shrink, and public confidence erodes as algorithms and corporate systems quietly set the terms of care.
AI is already capable. The systems continue to improve rapidly. The open question is whether doctors, educators, and health systems will redesign professional identity, training, and daily work quickly enough to meet the moment. The next generation of physicians will not be defined by the volume of knowledge they can hold in memory. They will be defined by how wisely they collaborate with—and when necessary, override—the intelligence now available at the bedside. The technology is ready. The harder test is whether we are.
About The Author
