用AI开启一线护理的下一个时代（英）

frontline nursing with AI A survey of more than 500 US nurses highlights growing AI adoption but pointsto the opportunity for clinical-care organizations to reimagine—not simplylayer technology onto—nursing work. by Gretchen Berlin and Mhoire Murphywith Beth Bravo and Stephanie Hammer AI has reached the nursing floor—but mostly as an add-on, not a disruptor. Despite growinguse, so far it has left the fundamentals of care delivery largely unchanged. As in ourearlierresearch, nurses continue to express strong belief in AI’s potential, but this conviction has nottranslated into widespread use. Our latest survey insights from more than 500 US nursessuggest the real transformation will come not from simply deploying more AI tools but fromclinical-care organizationsredesigning how nursing work actually gets done, end to end (seesidebar, “Research methodology”). Research methodology McKinsey has been conducting researchto help healthcare organizations better understand howemerging technologies, including AI, are shaping care delivery. This article highlights findings fromthe 2026 McKinsey Nursing AI Insights Survey, which explores nurses’ current use of AI, perceptionsof its impact, and expectations for its future role in clinical practice. These results are not intended tobe fully representative of all nurses; rather, they provide directional insights into experiences,attitudes, and adoption trends to help inform healthcare leaders and stakeholders as they integrate AIinto care delivery. The 2026 McKinsey Nursing AI Insights Survey included 521 frontline registered nurses across theUnited States and was in the field from February 17 to March 17, 2026. The2024 American NursesFoundation Nurses Survey, jointly deployed with McKinsey, included 7,200 frontline registerednurses across the United States and was in the field in March 2024. Both surveys were conductedonline. The sample for both surveys focused on nurses in roles that provide direct patient care acrossa range of settings, including hospital and acute care, ambulatory and outpatient care, long-term careand rehabilitation, behavioral health and substance use disorder, community health, and the USDepartment of Veterans Affairs and military health systems. Survey questions explored nurses’awareness, current usage, perceived benefits and risks, and expectations related to AI in clinicalworkflows and patient care. All answers were based on the experiences and perspectives of theindividual respondent. All questions were optional; therefore, the number of responses may vary byquestion. Clinical-care organizations across the United States have turned to AI as a potential solution to afamiliar set of challenges: persistent workforce shortages, rising patient acuity and demand, andgrowing administrative burden. AI is beginning to take hold: Nearly 65 percent of surveyednurses report using more AI tools today than a year ago, signaling strong momentum. Yet adoption remains uneven: 23 percent of nurses report no use, and only about 2 percent say AI isembedded in everything they do. The care setting also shapes adoption intensity. A higherproportion of nurse respondents from physician practices are superusers1(21 percent) comparedwith those in acute care hospitals (5 percent). This suggests that certain workflows, technologyinvestments, or organizational structures may be more conducive to AI integration and thatbroader adoption depends on system-level enablement and workflow integration, not individualinitiative alone. However, belief in AI’s potential is widespread. More than 80 percent say AI can help improvepatient care at least somewhat, including 16 percent who believe it can do so significantly. Thegap between belief and adoption underscores a broader opportunity: not simply to deploy AI butto reimagine how frontline nursing work is designed and delivered. If done well, AI could helpenhance the nursing experience and strengthen nurses’ ability to deliver high-quality patientcare. AI adoption is broad but fragmented Survey responses reveal that AI usage is not confined to a small group of enthusiastic adopters.Instead, usage is distributed across the workforce but typically at low to moderate levels (Exhibit1). More than three quarters of nurse respondents say they use AI in their daily work but atvarious levels of adoption. Only a small minority, roughly 10 percent, are superusers. Exhibit1 Despite limited depth of use, nurses say they believe in AI’s potential to help improve patientcare (Exhibit 2). This suggests that the primary constraint is not nurses’ willingness but likelythat clinical-care organizationsare yet to sufficiently embed AI to fundamentally transformworkflows. Exhibit2 Trust, not awareness, is the primary barrier Contrary to common assumptions, lack of awareness or training is not the dominant barrier to AIadoption among US nurses, according to our survey. In fact, lack of knowledge has droppedfrom the third-most-cited top three