The Next Doctor May Be an Autonomous AI

Medicine has reached a question that once belonged to science fiction: what happens when an AI can provide better care without a doctor in the loop? A paper in the Journal of the American Medical Association argues that this turning point may arrive sooner than the healthcare industry expects.

Its authors say AI is often better at doctoring than doctors, and their forecast reaches across the core work of medicine. The debate is no longer about whether machines can assist physicians. It is about whether autonomous systems will eventually deliver the best care on their own.

Five Tasks Could Redefine Medical Care

The paper, titled “Will Autonomous AI Exceed AI-Physicians as the Best Medical Care?”, reviews all published articles on AI in medicine since January 1, 2024. The authors conclude that medicine is rapidly approaching a transition point at which AI alone will exceed physicians and physician AI-hybrids at five fundamental medical tasks.

• Taking medical histories
• Establishing a diagnosis
• Identifying which tests are needed
• Prescribing treatment
• Managing chronic diseases

The paper makes a claim that challenges the basic model of modern healthcare: “Humans in the loop degrade AI performance.” In this view, trained clinicians may not always improve an AI system’s decisions. Human intervention could introduce errors into a process that machines handle with greater accuracy.

Ezekiel Emanuel, an author, medical ethicist, and oncologist, said he dismissed the idea for many years that AI could perform core medical functions. Vinod Khosla’s arguments changed that view. Khosla raised the question “Do We Need Doctors or Algorithms?” in a 2012 TechCrunch article and published a treatise on AI in medicine in 2016.

Emanuel eventually asked Khosla to collaborate, and Neal Khosla, Vinod Khosla’s son who runs Curai Health, joined the effort. Emanuel challenged the timeline by asking, “We’re making a prediction for four years from now. You don’t think that [AI is] going to be way more advanced than a doctor?”

The coauthors predict that, in many cases, a “superior autonomous AI” will likely surpass humans using AI by 2030. Vinod Khosla has also predicted that AI would do 85 percent of what doctors did by 2035.

The Human Doctor Still Has a Powerful Case

Those predictions have sparked concern about removing humans from clinical decisions. John Whyte, CEO of the American Medical Association, expressed concern about taking people out of the clinical loop, especially when medical choices affect patients in direct and personal ways.

Robert Wachter, head of medicine at UCSF, says “The argument they’re making is important.” He has outlined a future in which first-class medicine becomes a collaborative process between AI and doctors, while the masses in “economy class” mostly make do with AI alone.

Wachter concedes that “AI and humans together are better,” but he also emphasizes the irreplaceable human attributes of trained physicians. That tension sits at the center of the debate. AI may analyze information and recommend a path, yet the physician’s role includes qualities that cannot be reduced to a diagnostic output.

Richard Barton Brown describes AI as “a tool, not a replacement for humans, and it’s a tool with issues like any other tool.” The warning matters because AI is not a colleague or a superhuman entity. As one argument puts it: “No. No. NO. I know you really want AI to be human or superhuman or almost human or whatever — it’s exciting science fiction isn’t it? But no, it’s a machine not a colleague.”

Dr. Joseph Nathan says AI and robotics will fundamentally transform healthcare and redefine the physician’s role. Yet clinician time remains one of healthcare’s scarcest resources, and AI investments are not making clinical work significantly easier today.

Radiology Shows Both the Promise and the Risk

Radiology offers a clear view of how AI can transform a profession without replacing it. A pioneering AI scientist predicted that computers would replace human radiologists, and Hinton made a similar comment in 2016. Those predictions have not come true. AI has changed radiologists’ jobs, but it has not replaced them entirely.

Radiologists now work to create an ideal AI-human team, with AI supplementing human skills. Radiology AI relies mostly on traditional machine learning, not large language models. These systems can assist with image analysis and interpretation, seeing more than 99 percent of what humans can see and adding further insights.

That reach creates a new responsibility. Radiologists need to understand AI’s overall reliability while still spotting its mistakes, and some experts believe AI can help them detect subtle or unexpected details. At the same time, automation bias may cause clinicians to trust a machine’s answer without enough scrutiny.

Radiologists’ skills may also degrade over time as AI systems see more than 99 percent of what humans can see and gain additional capabilities. The models are trained using variant methodologies, creating another challenge for professionals who must learn when to follow a system and when to challenge it.

The pattern appears in colonoscopy as well. An analysis of 43 clinical trials found that “AI-assisted colonoscopies reveal more polyps than conventional ones.” Yet studies have also shown that doctors relying on AI for colonoscopy readings become less accurate in their own diagnoses.

That contradiction captures the future of medical AI. A system can find more while human judgment weakens through dependence. Ensemble reads from AI will eventually eclipse most of the value of traditional radiologists’ roles, according to one prediction, but patient-facing roles still require humans.

Digital plates have been in use for about two decades, showing that medical technology has long changed clinical work without erasing the people who use it. The next phase will push that partnership much further, from image interpretation to diagnosis, treatment, and chronic disease management.

The central question is no longer whether AI belongs in medicine. It is whether healthcare can build a system that captures AI’s precision without losing the judgment, responsibility, and human attributes that patients still need. By 2030, the answer may come from autonomous systems. Until then, physicians and AI will continue competing, collaborating, and reshaping what care means.

Stay connected via Google News

Follow us for the latest AI News