By JEFF GOLDSMITH
In his wonderful and pragmatic new book, A Giant Leap, Dr. Robert Wachter warns his professional colleagues that simply confiscating the potential administrative and clinical staff savings created by AI could create a whirlwind of negative consequences for healthcare companies.
Nowhere is the explosive potential for reaction to AI’s inroads into care delivery greater than in nursing, hospitals’ largest professional spending category. Hospitals employ more than 1.8 million registered nurses (RNs) and another 400,000 non-RN nursing staff. Registered nurses alone represent more than 30% of the hospital’s salaried workforce and more than 40% of overall staffing costs.
Nursing productivity is a central issue in overall hospital performance and a key intervening variable in both clinical quality and patient satisfaction. Therefore, the ability of AI to improve nursing productivity will be a central issue in determining the effect of AI on the hospital’s overall operational performance.
It is evident that there is room for improvement. Studies have shown that nurses spend only 25 to 30% of their work hours on direct patient care activities. The potential for AI to alleviate the enormous administrative burden that hurts nursing productivity could be the biggest benefit AI could provide. AI could materially increase nursing time at the bedside, increasing both patient and nursing satisfaction.
However, AI could also reduce hospital nursing staff, a factor that could, in turn, reduce nursing union membership, the largest and fastest-growing single category among hospital employee union membership. Nearly 18% of all registered nurses employed in hospitals are members of unions (AFSCME, AFT Healthcare, National Nurses Union, etc. and their local affiliates). Nurses’ union dues represent hundreds of millions of annual revenue for the unions that represent them.
The nursing unions’ most visible public policy initiative, which first appeared in California twenty years ago, was to get the state legislature to require nurse-to-patient ratios in hospitals. These were designed to force hospitals to hire more nurses with the intention of improving patient safety. What the ratios really did was throw more nursing corps into broken processes and systems. These laws had the important collateral benefit of ensuring a “guaranteed income” in union dues from more nurses employed in hospitals subject to these ratios!
Since then, formal (although less comprehensive) mandates on nurse staffing ratios have been extended to Oregon, Massachusetts, and New York, and legislation is pending in Maine, New Jersey, and Pennsylvania. Michigan, Minnesota and Washington State. Research on the anticipated qualitative benefits of California’s state-mandated indices confirms the expected benefits to patients, although the studies were based on correlational analyzes versus states without the indices mandate, not on pre-post studies of the indices’ effects on patient care.
Other studies concluded that the ratios increased both the number of registered nurses and compensation compared to other job categories, as well as hurting hospitals’ operating margins relative to states that lack the mandates. The point-counterpoint of these studies gives the impression that an issue is quickly becoming politicized.
AI joins other technology-enabled initiatives, such as telehealth-assisted virtual nursing, robotic medication dispensing, and “hospital at home” remote monitoring (enabling earlier patient discharge from the hospital setting), in threatening to undermine state-mandated nursing staffing ratios.
Nursing unions realized the threat of AI and raised the alarm. Consider the warning from the National Nurses Union: “The hospital industry, in cooperation with Silicon Valley and Wall Street, will use AI to further its dangerous effort to displace registered nurses from the physical care of their patients, prioritizing free or low-cost labor over patient needs.”
On the threat posed by remote monitoring, which will likely be structured and guided by AI, NNU warned: “This contributes to an ongoing effort by the hospital industry to maximize revenue by shifting care to less-skilled medical workers, or even non-medical workers in remote settings (e.g., the patient’s home). Over time, this will dramatically limit opportunities for nurses to care for patients in a hospital setting.”
Navigating this tense political and labor relations landscape will impose limitations on both the design and implementation of nursing-related AI applications. AI architects and their hospital administrative partners will win plaudits for streamlining medication administration and eliminating wasteful manipulation of the electronic health record (which consumes 30% or more of a nurse’s work hours). Both are sources of burnout and job dissatisfaction among nurses. Freeing up nursing time to spend on direct patient care will benefit both patients and caregivers.
Distrust of management and fear that management will simply pocket the savings from increased nursing productivity thanks to AI is what is driving union activism, hence Wachter’s warning. Conditions surrounding the implementation of AI in nursing can be expected to rise to the top in collective bargaining as contracts enter the renewal cycle.
In a recent essay on AI deployment, Stuart Winter Tear makes a crucial point about AI: “Agents are not being deployed. Work is being redesigned.” In this spirit, the heart of the matter is how the division between agentic action and human work is structured in dynamic work environments such as nursing.
This means that the participation of human caregivers in that redesign is crucial to the legitimacy of the effort. If nurses feel that their professional world is being reshaped by a cabal of (mostly male) AI technologists and finance people, the seeds of deep and bitter alienation will be sown. Relations between labor and management could deteriorate markedly, whether the facility is unionized or not.
How healthcare executives navigate the implementation of AI in nursing will be one of the most complex and fraught issues in care delivery in the coming years, amplified by the high level of anxiety about AI in society at large. As Wachter said in Giant Leap: “When doctors and nurses perceive that autonomous AI is safe for patients and can take onerous tasks off their shoulders, removing the doctor can be very easy. But if doctors perceive a threat to their income, status, or employment, expect a strong reaction.” The burden of proof regarding the contribution of AI to improving patient safety and the willingness to share power in implementing AI with direct care providers will fall squarely on the shoulders of management.
Jeff Goldsmith is a veteran healthcare futurist, president of Health Futures Inc and a regular contributor to THCB. This comes from your personal substack..
Acknowledgments: Bob Wachter, Bruce Vladek, and Trevor Goldsmith read this essay and provided constructive and helpful comments.