Healthcare workers warn: AI is no substitute for human caregivers

Earlier this year, the Cicero Institute, a conservative think tank, unveiled model legislation that would create a licensing pathway for artificial intelligence systems-an approach the institute promotes as a solution to healthcare worker shortages, gaps in access and rising costs.

But nurses and physicians say the proposal represents a fundamental shift in who-or what-delivers healthcare. They warn the proposal could pave the way for autonomous AI systems to take on responsibilities now performed by licensed clinicians, which would weaken patient protections and erode the human relationships central to quality care.

The Cicero Institute's AI Clinical Services Act would create a new licensing category for "AI Augmented and Autonomous Service Providers," allowing qualifying AI systems to replace, in whole or in part, licensed human clinicians under state oversight. Cicero's lobbyists have spoken in favor of the legislation to state lawmakers in Iowa and Idaho. Montana could become the next state on Cicero's list, and the Montana Nurses Association has the measure on its radar.

"Our concern is that this legislation treats AI as a complete substitute for the human foundation of healthcare. It would allow this limited technology to make decisions about patient care, eroding the expertise and dignity of healthcare workers, and give hospitals one more reason not to invest in safe staffing-especially in rural and marginalized communities. Patients have a right to compassionate, in-person care from qualified healthcare workers who know them, listen to them and can be held accountable for the decisions they make," said Vicky Byrd, RN, the CEO of the Montana Nurses Association and an AFT vice president.

One of the AFT's biggest concerns is the bill's creation of a path for AI systems-essentially algorithms-to replace licensed healthcare providers. Health professionals argue that patients need in-person care from licensed professionals, not technology offered as a substitute for them.

"AI does not have a place in healthcare when it comes to medical decision-making, and certainly not in front-facing patient communication, or as an autonomous information provider, or semi-autonomous medical professional," said Dr. Michael Massey, a second-year family medicine resident physician in Maryland and vice president of the University of Maryland Resident and Fellow Alliance, which represents over 950 resident and fellow physicians. Autonomous AI should never replace clinical judgment, said the Montana nurses. "Nursing is fundamentally a human profession built on assessment, advocacy, therapeutic communication, compassion, ethical decision-making and individualized care," said Byrd. "These cannot be replicated by technology."

The wrong answer to workforce shortages

The legislation's "expanded access" provision would allow autonomous AI systems to provide care in medically underserved communities under certain circumstances. Health professionals worry policymakers could rely on technology instead of investing in hospitals and healthcare workers-particularly in rural and frontier communities, where Montana's healthcare workforce crisis is already acute.

Massey said the proposal fails to properly understand the problem. "There is still a physician shortage in the U.S.," he said, noting that fewer physicians are entering primary and preventive care while more choose highly specialized fields. "I wholeheartedly disagree that AI would be a solution to any of the problems described as physician shortages. If hospital systems transition to AI providers, [for primary or preventive care visits], fewer and fewer people will choose to go into primary care," Massey said.

Montana nurses share that concern. They warn AI could become a lower-cost substitute for investing in healthcare professionals and rural hospitals at a time when many communities already struggle to maintain access to care.

Rather than replacing parts of the workforce with technology, they argue, hospitals should invest in recruiting and retaining nurses and clinicians through competitive pay, safe staffing and sustained support for rural and critical access hospitals.

Healthcare professionals also worry that overreliance on AI could erode the skills and independent judgment of nurses and other clinicians. If workers are increasingly expected to follow AI recommendations rather than make their own assessments, union leaders warn, clinical judgment could be stripped away.

The concern is heightened by the proposal's pathway for "fully autonomous AI" to perform tasks previously carried out by healthcare professionals, including diagnosis, triage, treatment and prescribing. Health professionals depend on their ability to notice subtle changes in a patient's condition, ask questions, interpret information in context and recognize when something is wrong-even when a chart or set of numbers does not tell the whole story.

Medicine depends on human relationships

For Massey, medicine is about far more than diagnosing illness. "It requires a human connection without exception," he said, noting that physicians spend much of their time educating patients, building trust and helping people understand complex medical decisions. "All of the art of medicine is persuasion and an ability to communicate what we know from our medical background to a patient that does not have that background," he said.

Massey described the conversations central to primary care: explaining why a patient should start a medication, helping someone understand what an elevated A1c means or working with a patient over time to stop smoking.

"An AI system can tell someone to stop smoking," he said. "But that is not the same as a healthcare professional looking a patient in the eyes, explaining the consequences and then following up weeks later to help the patient reach that goal."

Healthcare workers also worry the legislation could reduce patient choice by allowing AI systems to determine what care is in a patient's "best clinical interest" without meaningful collaboration between patients and their providers. Health professionals fear individualized care developed through shared decision-making could give way to standardized recommendations driven by algorithms. Massey also questions claims that AI would reduce healthcare costs.

He explained that patient visits are reimbursed based on what physicians and medical personnel enter as diagnosis codes, billing codes and other charge capture codes, and that AI systems will likely be designed by those who are most interested in generating a profit. "AI systems would be supercharged to get the most money out of every patient interaction," he said. "This would not lower healthcare costs at all," he said.

A debate about the future of care

Healthcare workers repeatedly raise another question: What happens when autonomous AI makes a mistake? Massey said the legislation leaves that unanswered. "It doesn't seem like there would be any accountability," he said.

Healthcare professionals are licensed by a state medical board, held to ethical standards and expected to learn from mistakes through established systems of accountability. Massey questioned whether autonomous AI systems can be held to those same standards.

Healthcare leaders say the potential consequences extend beyond individual patient care. They are looking closely at the legislation's "anti-protectionism" language, which they fear could be used to challenge safe staffing requirements and other regulations designed to protect patients and healthcare workers.

The proposal seeks to outlaw rules that protect the "professional boundaries, market share

or economic interests" of healthcare professions. Workers worry hospitals could use that language to argue that staffing requirements or other workforce protections improperly restrict AI deployment.

For healthcare workers, the debate is bigger than whether artificial intelligence belongs in medicine. The question is if, how and when technology should support healthcare professionals.

Massey recalled a patient whose dementia had progressed more significantly than previously recognized. During an in-person visit, he realized the patient had missed several appointments and needed closer follow-up. Massey arranged for the patient to return within weeks and began working to get him additional support. It is that kind of observation, follow-up and human responsibility that Massey fears could be lost if healthcare systems turn to autonomous AI as a substitute for health professionals.

"To exchange that safety net with an uncaring, profit-generating chatbot is not going to be addressing that issue at all," he said. "We are a safety net for people."

[Adrienne Coles]