NEW YORK, September 13 — The shift is coming to health care, and it may not look the way the doomsayers predicted. In a Perspective article published Sept. 12 in The New England Journal of Medicine, Dr. Dhruv Khullar argues that the adoption of artificial intelligence agents could lead to more, not fewer, health care professionals working in the United States. Khullar is an associate professor of population health sciences at Weill Cornell Medicine and a hospitalist at NewYork-Presbyterian/Weill Cornell Medical Center.
His article surveys economic theory and the history of technology in health care to push back against common predictions of AI-driven job loss.
The efficiency paradox
He does not deny that clinical roles will change, and he acknowledges that some types of health care jobs could be replaced. But he contends that economic theory and history offer distinct reasons to believe that in the long run, adoption of AI agents — even models capable of performing some forms of cognitive work — could lead to an expansion, rather than a contraction, of the clinical workforce. Cataract surgery and joint replacement are his examples.
Both procedures have become more efficient by reducing clinical effort, shortening recovery time and improving safety, which has enabled more patients to receive them. Khullar reasons that AI may have a similar effect on other health care services, especially if AI systems are priced near marginal cost and reduce the overall cost of delivering care.
Why the job-loss forecasts miss the mark
Expectations of AI-driven job loss, he argues, are influenced by the “lump of labor” fallacy, which holds that there is a fixed quantity of work to be done. Instead, he writes, the type and amount of work changes over time in response to technological advances, and AI may allow clinicians to prevent or treat conditions in ways that aren’t currently possible or imaginable. “AI may increase the demand for new professional capabilities by creating new treatments, care modes, and opportunities for specialization,” he writes.
Delivering high-quality care, Khullar notes, involves much more than successfully executing a given task. The practice of medicine depends on numerous skills and many interconnected steps, from interpreting test results to developing and negotiating treatment plans.
He invokes “O-ring theory,” named for the faulty components on the Space Shuttle Challenger, which contends that a single error at any step can seriously undermine outcomes. That may be especially relevant for many high-stakes situations in health care, which will continue to require the supervision of clinicians for reasons of safety and trust. All of this suggests that automating tasks doesn’t necessarily mean automating jobs.
“In fact, if AI automates some clinical tasks, the value of nonautomated, human tasks may increase,” Khullar said. The article was published by the Weill Cornell Medicine Office of External Affairs in New York.
Watch this space — the trend underneath is that the AI-and-jobs debate in medicine is being reframed, and the next round of evidence will come from how health systems actually deploy these agents.
