Healthcare looks to AI to alleviate burnout, labor shortages

Imagine being in a hospital where artificial intelligence (AI) is used in every step of your patient journey, from your admission to predicting when you might fall to even assisting doctors and nurses in your diagnosis and care. Meanwhile, for healthcare leaders and providers, this integration of AI may be the answer to delivering care more effectively amid persistent labor shortages and shifting reimbursement models.

Since May 2025, this scenario has been a reality at China’s Beijing Tsinghua Chang Gung Hospital, the teaching hospital associated with Tsinghua University. Earlier, in 2024, researchers at the university created the world’s first AI “Agent Hospital,” which consisted of 42 AI doctors operating across 21 departments capable of treating 10,000 patients in days. However, unlike the Agent Hospital, which was all AI and treated only simulated patients, the Beijing Tsinghua Chang Gung Hospital consists of human doctors, nurses and other staff using AI in their work caring for real patients.

In the U.S., the momentum behind AI adoption in healthcare also is unmistakable, though healthcare leaders note both the opportunities and challenges it presents. For instance, although 57% of C-suite healthcare executives in a 2025 Sage Growth Partners survey ranked AI-based clinical solutions as their top technology initiative over the next two years, only 13% said their organization has a clear strategy for integrating AI into clinical workflows. Meanwhile, 49% say appropriate use of AI is among their top three greatest challenges.

Jeffrey Covington, director of healthcare research and data analytics for BOK Financial®, attributes this push to adopt AI in the industry to its ability to remove administrative friction from clinical care. “It’s a huge time saver,” he said.

One of the clearest examples of AI’s time-saving capabilities is ambient documentation, where AI listens to the doctor-and-patient visit and translates it into structured clinical notes, Covington said. This differs from the previous practice of doctors dictating notes from a visit into a voice recorder afterwards. For physicians, the impact of ambient documentation can be profound: hours of evening documentation reduced, fewer backlogs and more time spent directly with patients.

A workforce stretched thin

The biggest force driving AI adoption in healthcare may be the labor shortage—of not just physicians, but also of Certified Nursing Assistants (CNAs), Registered Nurses (RNs), Licensed Practical Nurses (LPNs) and healthcare administrators. Compounding the problem, as Baby Boomers age, they’re requiring more and more care.

Altogether, this means that more healthcare staff is needed while job openings remain difficult to fill. Even with expansions in nursing programs and workforce pipelines, Covington sees no near‑term fix. “We’re in a situation where we still don’t have enough caregivers, and the training pipeline is not growing fast enough,” he said.

As in other industries, AI can help alleviate this problem by making the workers who are there more productive and have further reach. For instance, specialists who are difficult to hire in rural areas can support rural hospitals through tele‑ICU and remote monitoring, Covington said. The Rural Health Transformation Fund is already helping to fund these types of programs.

Additionally, remote monitoring tools—some already widely used—can detect fall‑risk behavior in skilled nursing facilities or track post‑operative patients through rehab protocols. This use of AI has the potential to increase worker safety and reduce the strain on direct-care staff.

AI is also reshaping follow‑up care and rehabilitation. Covington sees tremendous potential in using AI and monitoring tools to keep patients on track without requiring frequent in‑person visits, reducing the toll on patients while also helping facilities contend with limited staffing. “There’s a lot of different things with rehab that can be done with AI to improve patient outcomes,” he said.

He gave the example of cardiac rehab. Normally, after heart procedures like bypass surgery, patients have to go in person to a facility multiple times a week for rehab. With wearable-AI technology and remote-monitored rehab equipment, there’s the potential for rehab at home. Additionally, the data being collected from these devices are helping to further improve patient outcomes.

Legal and operational headwinds remain

Despite AI’s potential, Covington cautions that several obstacles remain. Malpractice risk is a major concern, especially as predictive tools move deeper into diagnostic support. He expects insurers to introduce exclusions or limitations on AI‑assisted care. “You’re going to need to start manuscripting these policies within alternative risk structures,” he said, urging health systems to work with risk managers early.

Data infrastructure is another hurdle, he said. Large volumes of proprietary clinical data must be stored, governed and accessed consistently for AI to be effective. Whether on‑premises or cloud‑based, these systems require investment and planning—especially for organizations still modernizing their technology stacks.

However, even with challenges ahead, Covington believes the direction is clear. AI is already reshaping documentation, care coordination, rehabilitation and the management of chronic and acute conditions. Its ability to extend the reach of clinicians and improve workflow efficiency offers tangible relief in a system long constrained by workforce shortages and administrative complexity.

He points to the use of AI in pre-authorizations. “Not all use cases will improve care and care delivery. Payors are already using AI to assist in pre-authorizations and denials. With providers becoming more adept at using AI to improve billing practices, revenue cycle managers are finding themselves on the frontlines of a new AI-enabled battlefield as both sides work to gain the upper hand.”

But he remains realistic: “These are highly complex processes requiring coordination and buy-in across departments, clinicians, regulators, business leaders and vendors. Adoption will vary across the country,” he said.