When Robots Play Doctor

Healthcare systems are adopting AI at warp speed, saving valuable time but potentially flubbing diagnoses. 

When Marcia visited urgent care with a broken bone, one AI agent scheduled her appointment, another transcribed her clinician’s notes, and a third read her X-ray. A fourth aligned her diagnoses with insurance-payment codes. What could possibly go wrong?

Perhaps nothing, perhaps plenty. Healthcare has long been known for painstakingly researching interventions and gauging patient outcomes before adopting new technologies. But in the case of AI, clinicians at many facilities have adopted the technology at warp speed, too fast for research on patient outcomes to keep pace. Few raise their eyebrows at the thought of AI navigating insurance payments. “But it can also get scary,” says Greg Button, president of global healthcare services at Korn Ferry. “Would you want to be diagnosed by a robot?”

To be sure, healthcare operations don’t rely on AI for final diagnoses. But the technology can certainly influence outcomes, and there is substantial evidence that it is commonly implemented in healthcare settings without significant research into its impact on patient outcomes. A 2025 study of 43 health systems found that 100% of respondents said they used ambient notetakers—yet only 19% of institutions reported high success in using AI for diagnosis, and just 38% have used it to successfully access patient risks. AI adoption is particularly messy in lower-resource settings. While well-funded health systems are more likely to customize predictive AI tools—for use in everything from billing to appointment scheduling—less-resourced institutions often adopt off-the-shelf versions, which may be designed for different patient populations, and can lead to ineffective or harmful outcomes.  

Even in the absence of established best practices, AI is still being adopted quickly in myriad healthcare contexts. Certainly, critical shortages among doctors and nurses are forcing some hospitals to deploy the new technology. “You can look at the entire value chain of healthcare and say that AI is playing a part everywhere,” says technology expert Maneesh Dube, senior client partner in Korn Ferry’s Executive Search practice. In pharmaceutical applications, AI can help identify the most effective molecule for combating a disease, while in X-rays, MRIs, and other tests, AI tools can catch abnormalities (human eyes are still also part of the process). On the one hand, AI in clinical settings removes variability from the process (which can reduce distractions that may have no bearing on the case); on the other, it eliminates the human perspective of a doctor who knows the patient’s family. “That’s both the benefit and the criticism,” says Dube.

Healthcare is unique in that the industry’s back-end systems have been notoriously clunky. There hasn’t been enough money to fund large-scale technological upgrades while also navigating a web of regulations. Other industries have far fewer restrictions. “The data in healthcare tends to be messy, siloed, and sensitive,” says Doug Greenberg, North America market leader for healthcare at Korn Ferry. This is a far cry from, say, mass-market retail, which rolls out AI agents and supply-chain robots as soon as they’re developed. “There’s obvious risk and regulation in healthcare,” he says, “so deployment is slower, more cautious, and more compliance driven.” Ideally, AI models, like medical devices, will be both testable and explainable. But today, for the first time in decades, healthcare is no longer considered the egregiously out-of-date industry. “We’re not decades behind, as we once were,” says Greenberg.

Learn more about Korn Ferry’s Total Rewards capabilities. 

Read more This Week in Leadership articles