A new paper has pushed the debate over medical AI into uncomfortable territory. Its authors argue that AI already surpasses doctors at several core tasks and could soon care for patients on its own better than a human physician.
The paper, published in the Journal of the American Medical Association in September 2026, says AI-alone care may outperform both physician-only care and a human-AI team when handling cognitive medical work. Ezekiel Emanuel, a bioethicist and one of the paper’s authors, put the prediction plainly: “As AI improves, having humans in the loop will likely worsen patient care.”
A bold prediction about doctors and AI
The paper says AI already beats doctors at five fundamental tasks. Those tasks include gathering relevant medical information from patients, diagnosing illness, conducting diagnostic testing, and managing chronic diseases. The paper does not treat AI as a tool that simply helps physicians work faster; it presents AI as a system that may perform the thinking at the center of medical care.
Its authors also argue that the gap will widen. AI is improving, while doctors’ abilities may diminish as they depend on machines for more of their work. That process, often called AI-deskilling, could leave physicians less practiced at remembering information, making judgments, and developing their own clinical skills.
Emanuel and venture capitalist Vinod Khosla, the paper’s authors, are making a prediction about the near future rather than describing a finished change in healthcare. “We’re making a prediction for four years from now,” Emanuel said.
That forecast has met resistance from the American Medical Association. John Whyte, the organization’s CEO, said the AMA sees promise in medical AI but does not support removing physicians from the care plan. “The AMA does see the potential in these tools, but they have to be utilized in the context of a care plan that’s governed by a physician,” Whyte said.
The practical gains and hidden costs
AI is already entering clinics in several ways. Around two thirds of physicians in the US use OpenEvidence, an AI chatbot designed for medical use. In the UK, up to four in 10 GPs use AI scribes that listen to consultations and create clinical notes.
Those tools can reduce administrative work. At a clinic in Dudley, an AI tool helped cut a backlog of patient letters from six months to 14 days. Recent months have also brought reports of doctors in the West Midlands using transcription tools, showing how quickly these systems are moving from experiments into daily routines.
Many clinicians welcome that change. Dr Lucas Seuren said, “Many clinicians are excited about ambient AI scribes, because they promise to cut down on paperwork.” Less paperwork can leave more room for a meaningful conversation with a patient, but the same system can create new risks.
AI scribes listen to audio, so they can miss physical gestures, facial expressions, and vocal tone. Those non-verbal cues can carry information that never appears in spoken words. AI summaries also tend to favor clinical information over patients’ stories and experiences.
Seuren warned, “The experiences of patients are poorly considered, and there are real risks that the patients’ stories are lost.” Patients may also share less sensitive information when they know a consultation is being recorded and processed by AI.
Why human oversight remains part of the argument
Using an AI scribe can create what researchers call “cognitive offloading.” The clinician no longer needs to remember every detail from a conversation, which may support a more focused exchange. But that convenience may also weaken memory recall and reduce opportunities to build clinical skills.
The concern goes beyond note-taking. A team of neurosurgeons in the UK used an AI tool to guide them during brain tumor surgery, showing that AI can support decisions in situations where the stakes are high. At the same time, doctors who rely on AI for more of their work may get less practice handling those decisions themselves.
This creates the central conflict in the debate. The JAMA paper argues that human involvement may eventually reduce the quality of care in cognitive medical tasks. The AMA argues that AI should operate within a care plan governed by a physician. Both sides recognize that AI can change medical work, but they disagree about how much authority doctors should keep.
The next stage of healthcare AI will not be judged only by whether a system can produce a diagnosis or a clean clinical note. It will also depend on whether the technology captures the full patient story, protects sensitive conversations, and helps clinicians without allowing essential skills to fade.
Based on
- American Medical Association Fires Back Against Paper Finding AI Is Now Outperforming Doctors — futurism.com
- Doctors using AI to take patients’ notes failing to capture ‘potentially vital information’ | The Independent — independent.co.uk
- Healthcare’s AI Debate Is Finally Moving Beyond Hype — forbes.com
- AI tools that take notes for GPs could miss vital information, study suggests — bbc.com
