As that workplace advocacy and bargaining continue, educators and researchers also have their eyes on the future: They’re trying to build solutions to support the next generation of nurses through training and involvement in how patient-facing AI is developed and deployed. By giving nurses a voice in AI’s inevitable disruption of health care, they hope to make the current adversarial relationship more collaborative.
Nurses’ perspective on the impact of clinical AI is often overlooked, said Jing Wang, dean at the Florida State University College of Nursing — despite being the “biggest users” of the technology. Of the 18 million people who work in health care, more than 4 million are nurses, compared to about a million doctors. But as clinical AI becomes commonplace in hospitals and home settings, they have often been left out of conversations about appropriate deployment of the tools.
“Nurses are fearful of AI,” Wang said, and “it’s because nurses were oftentimes not at the table when decisions were made to adopt AI.” While their logs of vital signs and medications have become the fuel that powered a new generation of automated sepsis alerts, for example, they are also the ones responsible for responding to frequent false positives and caring for patients the algorithm missed. “That’s where the tension is,” Wang said. “I had a bad experience with technology before that didn’t work. What do you expect me to do?”
Wang wants to help change those dynamics. In 2024, FSU launched the first nursing degree with a concentration on AI, with the goal of training AI-literate nurses who can think critically about the pros and cons of the rapidly evolving technology in their work.
“People feel like, well, these tools are going to change anyway, so we don’t need to teach them — leave that to the hospital,” Wang said. “That’s not right. We actually have to teach them whatever we have right now,” so they can develop the skills to evaluate new tools.”
It’s a struggle to find the right balance, Wang said. Right now, the master’s program with a concentration in AI is focused on AI application: Once you have a new tool, how do you think about implementing it and applying it in the clinical setting to support the patient, and what are the appropriate guardrails for use?
Toward the end of the program, each student spends a semester working with nurse preceptors to propose or implement an AI solution. Some students work with an IT department to vet several AI vendors, or with private companies testing telehealth and virtual coaching. Others will submit a tool for institutional review, including details of how it will be evaluated for performance, assessed for cybersecurity, and implemented in compliance with patient privacy protections.
With that kind of experience, the hope is that once they’re at the bedside, new nurses will be able to support the appropriate implementation of new tools when consulted, and raise evidence-based concerns when they’re necessary.
That’s the opposite of the dynamic some nurses see today, said Cathy Kennedy, a president of National Nurses United, which has conducted surveys of workforce opinions on AI and published a nurse and patient AI bill of rights, as they advocate for working nurses and their patients.
“The fear is when you have nurses that are new to their careers, they don’t have the lived experience” to question the output of less-than-perfect AI tools, and they’re more likely to simply go along with their suggestions, Kennedy said. “We’re trying to explain it’s important to use your critical-thinking skills.”
When nurses are unionized, they can use the power of collective bargaining to realize some of those changes. “If there’s a change to any working conditions, including an introduction of any technology, we have a right to bargain over those conditions,” Kennedy said. “Do they get around that? Absolutely. Do the nurses fight back in unionized facilities? Absolutely.”
In nonunion facilities, pushing back can be more difficult, contributing to the burnout that has pushed many nurses out of the workforce.
“Patients are not just as sick, but sicker than they’ve ever been before in those environments,” said Rae Walker, who directs the nursing PhD program at the UMass Amherst. “The nurses have as many, if not more, demands, and they’re being asked to do work that often exceeds human capacity.” Increasing adoption of AI often means that nurses are on the hook to constantly respond to continuous monitors and automated alerts.
To aid nurses in those challenging work conditions, Walker in 2025 published the Digital Defense Toolkit, a compilation of resources and tools to better equip nurses and other care workers to protect themselves and those they care for. The goal, Walker said, is “both to intervene in spaces where the AI technologies appear to perhaps be causing harm or posing a risk, but also where there is potential to actually intervene in a positive way.”
Among many potential steps, the toolkit suggests possible questions a nurse can ask about a tool: What recourse do I have if something goes wrong with an automated tool? Who is liable for any harms that result from the use of a technology? And if a tool is introduced as a way to improve efficiency and increase time spent with patients, what’s the plan to evaluate whether it’s fulfilling that objective?
Walker is also working to develop an index that helps health care workers measure how the design and implementation of digital tools are affecting their well-being and work environment. Often, the negative impact of an AI tool on the health care workforce isn’t as visible as potential patient harms, such as a false alarm or a missed prediction.
Algorithmic outputs can be used to determine how sick a patient is, for example, which informs how many nurses need to be on staff. “The tech is becoming more and more an excuse to reduce the degree to which there is a human workforce that is trained and consistent to be able to meet the needs,” Walker said. Nurse administrators are “in a sort of desperate situation right now, where they’re not the ultimate decision-makers, and they’re still pretty low on the budget power hierarchy, but they are nonetheless responsible for staffing their units.”
Advocates say that if new technology in medicine is meant to help existing clinicians provide better care, then actively involving nurses in the development of AI will help create solutions that are more likely to fulfill that promise.
At FSU, Wang and her colleagues this month received a $1.6 million grant from the National Institutes of Health to train nurse scientists to develop AI solutions and scientifically evaluate their ability to impact patient outcomes. And the school has already launched a consortium of AI developers, health systems, and educators to create guidelines for nurses’ participation in AI governance and adoption.
“We want more people to be generating the evidence, so nurses can be competent and confident in implementing those interventions that are powered by AI to guide their day-to-day practice,” Wang said. At the same time, she acknowledges that the technology is evolving more rapidly than anyone can keep up.
“It’s coming so fast,” NNU’s Kennedy said. “We really need to get a handle on this to make sure we’re doing right by patients.”