Health systems have spent years emphasizing the need to control costs, especially as reimbursement rates and inflation failed to keep pace with operational demands. In 2025, these cost-containment efforts will persist, and with an added challenge: the expenses of adopting AI tools and the "babysitters" needed to oversee them.
Such is the takeaway of an article from CBS News and KFF Health News, which checks in with numerous health system technology and AI leaders on the labor needs associated with AI investments. Right now it's a paradox: While AI tools are often touted as the future of efficiency and automation, they currently demand significant human oversight and attention to function properly.
"Everybody thinks that AI will help us with our access and capacity and improve care and so on," Nigam Shah, PhD, chief data scientist at Stanford Health Care, told the outlets. "All of that is nice and good, but if it increases the cost of care by 20%, is that viable?"
AI comes with its own set of challenges, including algorithm decay — where predictive accuracy diminishes over time. Integrating these tools into health system operations requires continuous monitoring and dedicated staffing to ensure fairness, reliability and effectiveness. At Stanford, Dr. Shah revealed that auditing two AI models for reliability took eight to 10 months and 115 hours of manual labor. The tricky part isn't determining whether the tools work — but whether they continue to work.
And then there is another part of the problem: many health systems lack the talent and infrastructure to meet these demands. The FDA noted this risk earlier in 2024, with FDA Commissioner Robert Califf, MD, expressing concern that health systems do not have the infrastructure and tools to make the most important determinations about whether an AI application is effective for health outcomes.
"I have looked far and wide," Dr. Califf said at a recent agency panel on AI. "I do not believe there's a single health system, in the United States, that's capable of validating an AI algorithm that's put into place in a clinical care system."
Health system leaders have also expressed healthy skepticism about how much transformation is possible when AI solutions are layered on top of fragmented or flawed systems that are widespread in healthcare.
Read more about the paradox of healthcare AI and the need for oversight in the CBS News report here.