There are roughly four times as many nurses in American healthcare as physicians. When a tool changes the work at the bedside, nursing absorbs most of the change.
So it is worth noticing that nursing is consistently the last group asked.
The 2026 adoption data makes the pattern hard to miss. Nurses use AI tools at meaningfully lower rates than physicians, report far less employer guidance, and are much more likely to say the tools saved them no time at all. The instinct is to read that as reluctance. The data says something else: where nurses were actually trained, the time savings appear. Where they were not, the tools mostly generate extra steps.
That is not a technology adoption curve. That is an implementation failure with a predictable shape.
We have spent nine issues looking at tools arriving on hospital floors. This week we look at who was in the room when they were chosen, and who was not.
Let's get into it.
— Troy, Ray, and Ibrahim
AI is being deployed around nurses, not with them
Survey work published through 2026 puts numbers on something nursing has been saying anecdotally for two years. About 41% of nurses report using AI tools at work, against 57% of physicians. Nurse adoption nearly tripled in a single year, so the gap is not disinterest. When narrowed to clinical-specific AI tools used frequently or always, the figures are 30% of nurses and 37% of physicians.
The support picture is starker. Only 8% of nurses report a clear AI strategy from their employer, and reporting on the 2026 State of Nursing survey describes deployment happening largely around nurses rather than with them.
Why it matters: The training variable does real work here. Almost half of nurses using AI report little or no time saved. But among nurses whose employers provided thoughtful AI training, 24% report saving more than an hour a day, against 16% of those without it. Same tools, different preparation, materially different return. If your rollout is underperforming, the survey suggests looking at the implementation before blaming the software or the staff.
The catch: These are self-reported cross-sectional surveys, not audit-log studies, and perceived time savings are exactly the measure that ambient documentation research has shown can drift from objective reality. There is also a real substantive concern underneath the adoption numbers that training will not resolve: a large majority of nurses surveyed worry AI could erode critical thinking, and many fear displacement. Those are not irrational responses to a tool arriving without consultation.
Bottom line: The bottleneck in nursing AI is not nurse enthusiasm. It is that almost nobody is being trained, and almost nobody is being asked. Both are cheaper to fix than the software was to buy.
Read the full report from Incredible Health →
Where do you land on this one? Hit reply and tell us. We read every response, and the best ones shape the next issue.
From the Field
Nurses, this one is yours. Was your unit consulted before the last tool landed, or did you find out from a tip sheet taped to the workstation? What is the documentation or alerting change that made your shift genuinely better, and what is the one that just added clicks? We are also looking for the informatics nurses doing translation work between vendors and the floor, because that job is invisible and load-bearing. We protect your anonymity and your patients' privacy, always.
The Profile: Molly K. McCarthy, MBA, BSN, RN-NI
Howley Family Visiting Professor, Villanova University M. Louise Fitzpatrick College of Nursing; former U.S. Chief Nursing Officer, Microsoft
Molly McCarthy's career traces the exact path this issue is about. She started at the bedside, in the NICU and as a pediatric kidney transplant team coordinator at Stanford Medicine Children's Health, then moved into industry, eventually becoming the U.S. Chief Nursing Officer at Microsoft.
That trajectory matters because the translation problem in nursing informatics is bidirectional. Vendors build without understanding the floor, and floors receive without a voice in the build. People who have credibly stood in both rooms are rare, and nursing has spent two decades short of them.
- Served as U.S. Chief Nursing Officer for Microsoft, bringing clinical nursing perspective into enterprise health technology strategy
- Began in direct patient care in the NICU and as a Pediatric Kidney Transplant Team Coordinator at Stanford Medicine Children's Health
- Held roles in clinical informatics at Philips' Patient Care and Clinical Informatics division and in product marketing at Natus Medical
- Named Howley Family Visiting Professor in Healthcare at Villanova's College of Nursing for the 2025-26 academic year, focused on health strategy and technology
Read more at Villanova University →
Quick Hits
Nurses are using tools they were never trained on
Survey reporting in 2026 found nurses being asked to use AI tools they neither trust nor received training for, with most saying they feel unsupported. The same nurses rate AI's potential impact on their work optimistically. Wanting the tool and being set up to fail with it are not contradictory positions. LEARN MORE
Informatics nurses are the bridge, and there are not enough
HIMSS workforce data cited through 2026 shows demand for nursing informatics specialists growing sharply, with Chief Nursing Informatics Officer roles expanding fastest. These are the people who translate between the vendor demo and the actual workflow. Systems that skip the role tend to discover why about six months into a rollout. LEARN MORE
What nursing informatics leaders actually want
A qualitative study gathered nursing informatics leaders' perspectives on the path forward with AI in nursing, organized around education, empowerment, and elevating nursing voices. The recommendations are unglamorous and consistent: involve nurses in design, build literacy before deployment, and stop treating the largest clinical workforce as an implementation detail. LEARN MORE
Staffing remains the thing AI cannot touch
Across specialties and countries, staffing is the single most documented driver of nurse burnout, and unsafe ratios degrade assessment, communication, and error rates in ways no documentation tool addresses. Worth holding next to any vendor claim about reducing nurse burden. Software can move paperwork. It cannot staff a unit. LEARN MORE
Off the Clock: who gets asked
There is a quiet hierarchy in how healthcare technology gets decided, and everyone who works in a hospital can recite it without being told.
The vendor talks to the executive. The executive talks to the CMIO. The CMIO talks to a physician committee. Somewhere near the end, if the timeline holds, somebody builds a training module for nursing. The module is twenty minutes long and goes live the week before launch.
That sequence is not malice. It is the ordinary gravity of how institutions allocate attention. But it produces a specific, repeating result: the group that will touch the tool most is consulted least, and then gets described as resistant when the rollout underperforms.
The 8% figure is the whole argument in one number. Fewer than one in twelve nurses can identify a clear AI strategy from the organization deploying AI onto their shift.
What makes this more than a fairness complaint is that nursing has always been healthcare's sensor network. Nurses are the continuous observation layer. They are in the room when the number is normal and the patient is not. Every experienced physician has a story about being saved by a nurse who said something is off and could not yet say why.
That capacity is precisely what a poorly implemented alerting system degrades. Flood the sensor network with noise and you do not just annoy people. You reduce the signal that the entire safety architecture of a hospital quietly runs on.
So the case for bringing nurses in early is not primarily about respect, though it should be. It is that they are the ones who will notice first when a tool is wrong, and the only ones positioned to tell you before it matters.
It costs very little to ask them. It has cost a great deal not to.
Until next Sunday.
For the people keeping medicine human.
