AI May Expand Healthcare Workforce, Not Replace It

Economic forces may drive growth, not contraction

Artificial intelligence tools are rapidly entering clinical practice, raising widespread concerns about job displacement in radiology, primary care, pathology, and psychiatry. But a new analysis challenges the assumption that automation necessarily means fewer healthcare jobs.

Dr. Dhruv Khullar, an associate professor at Weill Cornell Medicine, argues in The New England Journal of Medicine that economic principles and historical patterns point toward workforce expansion rather than contraction as AI becomes embedded in clinical workflows. While acknowledging that specific roles will transform and some positions may disappear, Khullar contends the overall trajectory favors growth.

Why it matters

Healthcare leaders planning workforce strategy need frameworks beyond simple substitution models. If AI follows historical patterns of medical technology adoption, organizations should prepare for increased service volume and new specializations rather than downsizing—a fundamentally different strategic posture with implications for hiring, training, and capital allocation.

Three economic principles suggest expansion

Khullar's argument rests on established economic concepts rarely applied to healthcare AI discussions.

First, he points to Jevons paradox, where efficiency gains increase total consumption of a resource. Cataract surgery and joint replacement procedures became more efficient through technological advances—shorter procedures, faster recovery, improved safety—which enabled far more patients to receive them. If AI reduces the cost and effort required to deliver diagnostic imaging or primary care consultations, demand for those services may surge beyond current capacity.

Second, the "lump of labor" fallacy assumes a fixed amount of work to be done. Khullar argues this misunderstands how technology creates entirely new categories of work. AI may enable prevention and treatment approaches that don't currently exist, generating demand for new clinical capabilities and specializations that expand rather than replace the workforce.

Third, "O-ring theory"—named for the Challenger disaster component—recognizes that complex systems fail when any single element fails. Healthcare delivery involves numerous interconnected steps where human judgment, supervision, and coordination remain essential for safety and trust. Automating individual tasks doesn't eliminate the need for clinicians to oversee the entire process.

Tasks versus jobs

The distinction between automating tasks and automating jobs proves central to Khullar's thesis. Medicine requires interpreting results, developing treatment plans, negotiating with patients, and coordinating care across multiple touchpoints. When AI handles routine components, the value of distinctly human skills—judgment, communication, empathy, oversight—may actually increase.

This analysis doesn't dismiss disruption. Clinical roles will change substantially, and some job categories will contract or disappear. But the net effect, Khullar suggests, may resemble previous waves of medical technology that ultimately expanded rather than contracted the healthcare workforce.

The details were first reported by Weill Cornell Medicine in coverage of Dr. Khullar's Perspective published August 29 in The New England Journal of Medicine.