What artificial intelligence in nursing means for faculty

In a 2025 survey of 99 undergraduate nursing students in New York City, 92% said they used generative AI tools: to clarify nursing concepts, look up information, work through health assessments and get schoolwork done. That’s what artificial intelligence in nursing looks like from inside most programs today: software that drafts text, predicts risk and, more and more often, plays the patient in a simulation, already in students’ hands while the faculty is still deciding what to allow.

The same students listed the downsides themselves: subscription costs, over-reliance, information overload and answers they couldn’t fully trust. What they asked their school for was early guidance and short instructional modules, which is a sensible place for a faculty to start.

Outside the classroom the uses are wider. A systematic review of 21 studies by nursing researchers at the University of Almería grouped them into three families: early disease detection and clinical decision-making, support systems for patient monitoring and workflow, and nursing training and education. A 2025 umbrella review of 18 reviews in the Journal of Medical Internet Research reached a conclusion that lands squarely on schools: nursing curricula need updating with AI-driven educational tools, with AI literacy and ethical training built in alongside them.

The numbers point the other way. The WHO’s State of the World’s Nursing 2025 counts 29.8 million nurses in 2023, up from 27.9 million in 2018, and a global shortage that still stands at 5.8 million. On current trends it falls to 4.1 million by 2030. The profession’s problem, on the WHO’s own figures, is too few nurses.

The American Nurses Association is just as direct. Its 2022 position statement on AI says it in one sentence: “AI does not replace a nurse’s decision-making, judgment, critical thinking, or assessment skills.” It also quotes the Code of Ethics on the point that matters most for teaching: nurses “are accountable for their practice even in instances of system or technology failure.”

For a nursing program, that last line is the practical one. If the nurse answers for the decision whatever the software suggested, then checking what the software says becomes part of clinical judgment, and it has to be practiced like the rest of it.

Put together, those reviews describe four kinds of tools that nurses and nursing students run into today:

• Decision support and early detection. Models that analyze patient data to flag risk or deterioration earlier and support clinical decisions.
• Monitoring and workflow. Systems that watch patients remotely or help organize the work of a unit.
• Documentation. Tools that automate part of the charting. An umbrella review in Nursing Inquiry lists it among the efficiency gains, next to earlier detection.
• Education. Virtual patients and simulators, plus the general-purpose chatbots students already study with.

In nursing education the evidence is young, but there’s already enough of it to pool. A 2026 meta-analysis of 61 studies on AI in nursing simulation found improvements in students’ knowledge and in their skills, and its map of the field named generative AI, virtual patients and geriatric care as the themes gaining ground. A smaller scoping review of 11 studies recorded the costs as well: simulation-related anxiety, misuse, and concerns about data quality, privacy and academic integrity.

For a curriculum committee, “AI” is too big to decide on in one go. It helps to take training apart and look at each piece on its own:

The last row has a piece of its own: AI in medical education: a faculty’s playbook works through assessment, teaching, content authoring and the written policy one layer at a time, and most of it carries over to a nursing program as it is.

The first row of that table is the one most programs can’t staff. Standardized patients take budget and scheduling, so encounters per student stay few, and the conversation that needs the most repetitions gets the fewest. That’s the gap a virtual patient simulator is built to fill.

What happens under the hood, from what the virtual patient remembers to how a session gets scored, is covered in how an AI patient simulator works.

The umbrella review cited at the top also names the barriers programs keep running into: resistance to adoption, the lack of any standardized AI education, and unequal access to the technology. None of that has a settled answer yet, and programs are working through it in different orders.

So here’s a question for your next curriculum meeting: in which row of that table are your students already using AI, with no rule from the program yet?