AI Sparks Debate: Are Doctors' Days Numbered?
A new JAMA paper argues that autonomous AI will soon outperform both doctors and doctor-AI teams in delivering medical care, prompting pushback from physician groups over the loss of human oversight.
A paper published this month in the Journal of the American Medical Association argues that standalone artificial intelligence will soon deliver better medical care than either physicians alone or physicians working alongside AI. The authors, including oncologist Ezekiel Emanuel and venture capitalist Vinod Khosla, reviewed studies on AI in medicine published since January 2024 and concluded that AI is approaching a tipping point where it will surpass humans at five core tasks: taking medical histories, diagnosing, ordering tests, prescribing treatment, and managing chronic disease. They go further, suggesting that human involvement can actually degrade AI's performance.
Emanuel says he long dismissed Khosla's prediction that AI would handle 85 percent of doctors' work by 2035, a view Khosla has promoted since at least 2012. Emanuel changed his mind after reading a book by his colleague Robert Wachter of UCSF, which envisioned a future where wealthier patients get combined AI-doctor care while others rely on AI alone. Khosla's son Neal, who runs the AI-driven care company Curai Health, also contributed to the paper; the authors disclosed these financial ties. The team predicts that a "superior autonomous AI" could outperform human-AI teams by 2030.
Pushback from physicians
John Whyte, CEO of the American Medical Association, criticized the paper, noting that some of the reviewed studies were simulations rather than real-world trials, and not all supported the authors' conclusions. He cited a February 2026 Nature study showing that most patients struggled to effectively converse with large language models to extract useful medical guidance. Whyte insists AI tools must operate within a care plan overseen by a physician.
Wachter, despite being singled out in the paper, acknowledged its argument has merit but maintains that human physicians remain essential for tasks like delivering difficult diagnoses and guiding treatment decisions. He also warned of a "de-skilling" risk, where new doctors trained alongside AI may never fully master basic clinical skills. Vinod Khosla believes doctors will still be needed for surgery and emergency care in the near term, but says their role as medical experts is largely becoming obsolete, while Neal Khosla predicts AI will eventually gain regulatory approval to prescribe medication directly.
