A year ago, 15% of American nurses used AI at work. Now it's 44%, according to a national survey released July 7, 2026. That's not gradual adoption. That's a technology going from a minority habit to something nearly half the profession touches on an ordinary shift, in twelve months.

The numbers come from Incredible Health's seventh annual State of Nursing report, drawn from a survey of 2,240 US nurses and technicians combined with platform data from 1.5 million healthcare workers on the company's job marketplace. The headline growth number is striking on its own. What makes the report worth reading past the first paragraph is everything it found sitting underneath that growth: a workforce racing ahead of the institutions that employ it, using tools nobody trained them on, deployed under strategies almost nobody wrote down.

A Number That Tripled Without a Plan Behind It

Only 8% of nurses say their employer has spelled out a clear strategy for how AI should be used in their work. One in five say AI tools simply appear in their workflow with no explanation attached. Nearly half, 46%, report receiving no AI training whatsoever over the past year, even as usage nearly tripled across the same period.

That gap between adoption speed and institutional readiness is the actual story here, more than the 44% figure itself. Healthcare futurist Jeff Goldsmith, commenting on the report's findings, described the coming years as some of the most complex and fraught in care delivery, driven by exactly this collision between AI rollout and workforce readiness. Hospitals aren't resisting AI. They're deploying it faster than they're explaining it, and nurses are the ones absorbing that gap in real time, at the bedside.

Why Trust Hasn't Caught Up With Adoption

Separate reporting on the same survey, published by the Washington Post, found that more than 80% of nurses who use AI say the technology isn't accurate enough to rely on without double-checking it themselves. That's not a fringe skepticism. It's the dominant view among the people actually using these tools daily. Nurses aren't rejecting AI outright. They're using it and still verifying almost everything it produces, which raises an uncomfortable question about how much time these tools are actually saving if every output needs a human check anyway.

The survey data backs that up directly. Nearly half of nurses who use AI regularly say the tools have saved them little or no time in their daily workflow. Only 19% report saving more than an hour a day. Meanwhile, opinions split almost exactly down the middle on whether AI can meaningfully reduce documentation burden, one of nursing's most cited pain points: 36% see real potential there, and an equal 36% see no meaningful opportunity at all.

The Training Gap That Explains the Time-Savings Gap

One data point in the report cuts through a lot of the ambiguity. Among nurses who received thoughtful training from their employer on how to use AI tools, 24% report saving more than an hour a day. Among nurses who didn't get that training, only 16% report the same result. That's not a small difference, and it suggests the technology itself isn't the primary variable determining whether AI actually helps a nurse get through a shift faster. How well the rollout was handled is.

That finding puts real weight behind the report's central complaint: hospitals are buying and deploying AI faster than they're teaching people to use it well, and the resulting difference in outcomes isn't subtle.

Whose Input Gets Sought Decides Who Adopts

The report also found a clear link between being consulted on tool selection and actually using the tool afterward. Among nurses who helped choose the AI systems their workplace adopted, 81% reported using them. Among nurses who were never consulted, that number drops to 62%. Involvement in the decision correlates directly with willingness to use the result, which isn't a surprising finding in the abstract, but it's a specific, measurable version of a pattern hospital administrators clearly aren't acting on consistently yet, given how many nurses report tools simply showing up unexplained.

Fear of job displacement tracks a similar pattern. Among nurses who use AI regularly, 37% say they're less worried about AI taking their job than they were a year ago. Among nurses who only watch colleagues use AI without using it themselves, just 12% feel that same relief, and nearly half report feeling more anxious about job security than they did twelve months ago. Direct experience with the technology appears to reduce fear of it. Distance from it does the opposite.

An Automated Job Search Meeting a Manual Hiring Process

One of the report's sharper findings involves a mismatch nobody seems to be addressing. Thirty-nine percent of nurses have used AI in their own job search, a figure that rises to 60% among nurses under 28. On the other side of that same hiring process, only 4% of healthcare employers use AI for interviewing or hiring decisions. Candidates are optimizing applications with AI at a much faster rate than employers are adapting how they evaluate those applications, which is its own quiet mismatch playing out across an industry already struggling with retention.

Among the specific tools nurses report using for healthcare-specific tasks, Epic's built-in AI leads at 28%, followed by ambient documentation tools at 8% and OpenEvidence at 4%. That distribution suggests most AI use in nursing right now is happening through whatever's already embedded in existing hospital software, rather than through standalone tools nurses sought out independently, which tracks with the report's broader picture of a workforce adapting to what gets handed to them rather than choosing it themselves.

The throughline across every number in this report is the same: nurses are adopting AI faster than anyone is preparing them to use it well, and the resulting gap between usage and trust isn't a temporary transition phase. It's what happens when a technology gets deployed ahead of the training, strategy, and consultation that would make it actually work.