The nurse educator shortage is a time problem: what AI can and cannot take off faculty

The nurse educator shortage is usually described as a hiring problem, and it is one. It is also, more immediately, a time problem: the faculty a program already employs spend a large share of their week on work that does not require their clinical expertise, while the work that does require it goes undone.

The size and shape of the gap

The numbers are unusually consistent year over year, which is what makes them worth planning against. For 2025 to 2026, AACN data from 863 surveyed schools identified 1,588 vacant full-time faculty positions, a 7.2 percent vacancy rate. Adjusted for the additional positions schools said they needed to meet student demand, the figure rises to 9.6 percent. In the same period, 93,176 qualified applicants were turned away, with insufficient faculty, clinical sites, classroom space, preceptors, and budget named as the reasons.

The structural detail that matters most is the credential requirement: 80.9 percent of those vacancies require or prefer a doctoral degree. That narrows the hiring pool to a population that takes years to produce and that competes with clinical practice on salary. A program cannot hire its way out of this within a planning cycle, however well funded it is, because the candidates do not exist in the year the seats are needed.

Each unfilled position is generally estimated to cost a program eight to ten admitted students. That conversion is what turns a staffing statistic into an enrolment one, and it is the reason faculty time is properly treated as capacity infrastructure rather than as an HR line.

Where faculty time actually goes

Ask a nurse educator to account for a week and the pattern is familiar: preparation and content maintenance, grading and written feedback, clinical supervision and site coordination, student remediation, committee and accreditation work, and, if any remains, scholarship. The scarce and irreplaceable part of that list is narrow. Clinical supervision, evaluative judgement, debriefing, and mentorship require the expertise the program hired. Most of the rest requires a competent professional but not necessarily a doctorally prepared clinician.

This is where the honest version of the AI argument sits. The claim that AI will relieve the faculty shortage by teaching students is not credible and is not supported by any accreditation standard. The claim that AI can absorb a meaningful share of the non-substitutable work, returning hours to the work only faculty can do, is both credible and measurable.

An honest division of labour

The row worth dwelling on is remediation. In most programs, identifying a struggling student is itself expensive: it depends on a faculty member noticing a pattern across assignments, often several weeks after the pattern started. Continuous practice data moves that detection earlier and makes it specific, so the faculty hour is spent on the intervention rather than on the diagnosis. The intervention still needs the educator. The search for who needs one does not.

Why grading is the wrong place to start

When a program does decide to deploy AI against faculty workload, grading is the first candidate proposed and close to the worst one to start with. It has the least favourable ratio of benefit to exposure in the whole list. Summative grades are contested by students, reviewed on appeal, and inspected at accreditation, so an error there is expensive in a way that an error in a practice scenario is not. Grading is also the activity most entangled with professional judgement: the reason a submission fails is frequently the thing the student most needs explained, and that explanation is the teaching.

Practice and formative feedback invert every one of those properties. The stakes on any single interaction are low, errors are visible and correctable inside the session, and the volume is genuinely beyond human capacity, which is precisely where automation earns its place. A student can attempt the same deteriorating-patient scenario nine times in a week and receive immediate feedback on each attempt. No faculty member has ever been able to offer that, so nothing is being displaced from a person; capacity is being created that did not previously exist.

There is a sequencing argument here as well. A program that begins with practice accumulates evidence about how the tool behaves, where it is wrong, and how students respond, before anything consequential rests on it. A program that begins with grading learns the same lessons through appeals.

The preceptor half of the problem

Faculty vacancies dominate the discussion, but preceptor availability appears in the same list of reasons programs turn students away, and it behaves differently. A preceptor is not employed by the program. They are a working clinician who supervises a student on top of a full patient load, usually for little or no compensation, and their willingness is the single least controllable input in the whole system. Programs compete for the same finite pool, and a preceptor who has a poor experience with one student is measurably harder to recruit for the next.

That makes preceptor time worth protecting in a way faculty time already is. A meaningful share of a preceptor's first days with a student goes to orientation and to basic procedural rehearsal the student could have completed beforehand. When a student arrives having already run the common scenarios, having been assessed on them, and with a record the preceptor can actually read, the relationship starts at a more advanced point. The preceptor spends their scarce attention on the judgement that only a real clinical environment can teach, which is both a better use of the placement and a materially better experience for the clinician deciding whether to take another student next term.

This effect is easy to dismiss as soft, and it is harder to quantify than a faculty hour. It is also the mechanism by which a program either keeps its placement relationships or slowly loses them, which makes it a capacity variable whether or not it appears on a budget line.

The failure mode: measuring the wrong thing

Programs that adopt AI tooling and report no benefit usually made one of two errors. The first is measuring adoption instead of displacement: counting logins and completed activities, which says nothing about whether any faculty hour moved. The second is layering the tool on top of existing work rather than substituting for part of it, so students now do the practice module and the worksheet, and faculty review both.

The measurement that answers the question is narrow and boring. Which specific activity did this replace? How many faculty hours did that activity consume last term? Where did those hours go this term? A program that cannot answer the first question has not deployed a capacity instrument; it has added an activity.

What this does not solve

Two constraints are untouched by any of this, and claiming otherwise damages the credibility of the parts that hold up. Clinical placement seats are negotiated with health authorities and employers, and simulation affects them only to the extent a regulator permits substitution or a program redirects the repetition consuming those seats. The rules and caps that govern that are set out at does simulation count toward clinical hours.

The second is the doctoral pipeline. Nothing in a software deployment produces doctorally prepared faculty. What it can do is reduce the cost of each vacancy while it stays open, which is a different and smaller claim than the one usually made, and the only one a program should plan against.

Where to start

Begin with the single highest-volume repetitive activity in one course, not with a division-wide deployment. Establish what it currently costs in faculty hours, substitute it rather than supplement it, and measure where the hours went. One course, one term, one number produces an argument that survives a budget conversation. A pilot spread thinly across six courses produces enthusiasm and no evidence.

The capacity argument this sits inside is at training capacity is the metric that matters, and its extension beyond nursing to every allied-health program sharing the same faculty and placement pool at capacity problem beyond nursing. To model the faculty and placement hours a specific division could redirect, see the ROI estimator. To run the scoped version described above, the pilot process sets out how we structure it.