SAFE
Nurse practitioners physically examine patients, perform procedures like suturing, joint injections, and pelvic exams, and hold prescriptive authority that requires a licensed human signature — none of which an AI can execute or legally own. The documentation-heavy layer of the job (note writing, coding, prior-auth letters, patient education handouts, differential-diagnosis brainstorming) is exactly what current models do well, so ambient AI scribes and decision support will compress the clerical share of the day rather than the clinical one. The likely outcome is more patients per NP and faster panels, not fewer NPs; note that the licensure/prescribing shield is regulatory and scope-of-practice rules can shift.
Headcount grew steadily across the period.
Median pay $109,820 → $132,300 -3.6% in real terms
This line is counted by the Bureau of Labor Statistics — the one figure on this page that isn't a judgement of ours. Headcount moves on demand, offshoring, demographics and the business cycle, and automation is one term among several, often not the loudest.
So a falling line is not evidence that AI did it, and a rising one is not evidence that it won't. Both happen in this register: some occupations resist automation and shrink anyway, others are highly automatable and keep growing. The marked year is 2020.
BLS projection, 2024–2034
+40.1% 320,400 → 448,800 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +40.1% more of these jobs by 2034. Note that safe does not mean well paid — several of the fastest-growing resistant occupations are among the lowest paid on the register.
Different clocks. The score is what current AI could do to this work today. The projection is how many of these jobs will exist in 2034. Everything between the two — how fast employers actually adopt, whether demand grows in the meantime — is why they can point opposite ways without either being wrong.
~29,500 openings a year on average, including replacing people who leave.
Medical Surgery NurseNurse Practitioner (NP)Adult Nurse PractitionerAdvanced Practice ProviderSurgical Nurse PractitionerCertified Nurse PractitionerAdvanced Practice Nurse (APN)Cardiology Nurse PractitionerOrthopedic Nurse PractitionerDermatology Nurse PractitionerPsychiatric Nurse PractitionerFamily Nurse Practitioner (FNP)Family Health Nurse PractitionerNeurosurgical Nurse PractitionerGerontological Nurse PractitionerLicensed Nurse Practitioner (LNP)Neonatal Nurse Practitioner (NNP)Palliative Care Nurse PractitionerPediatric Nurse Practitioner (PNP)Gastroenterology Nurse PractitionerACNP (Acute Care Nurse Practitioner)Electrophysiology Nurse PractitionerInternal Medicine Nurse PractitionerEmergency Medicine Nurse Practitioner
Holding it up: liability shield . Weakest point: task resistance .
Mixed — a routine tier and a judgment tier A 13 reflects the split day: palpating an abdomen, draining an abscess, inserting an IUD, and reading a patient's affect across the exam table stay off-limits to software, but chart review, ICD-10/CPT selection, refill triage, standing-order protocol visits for strep or UTI, and the entire note-writing load are already being handed to scribes and algorithms — enough of the shift is screen work to keep this out of the 14+ band.
Hands-on in uncontrolled environments A 16 is earned in acute and primary care rooms where you do the pelvic exam, the joint injection, the wound closure, the neuro exam, and the code — uncontrolled bodies, uncooperative kids, home visits and SNF rounds — though it stops short of 18-20 because a meaningful and growing share of NP visits are telehealth and e-consult where nobody is touched.
Licensed human required and personally liable An 18 tracks the fact that your NPI and DEA number go on the prescription, your state licence is what a board can suspend, and in 27-plus full-practice-authority states you are the named defendant rather than a collaborating physician's extension — the two points held back acknowledge that scope-of-practice is statute, and legislatures rewrite statute.
Exists to be accountable for ambiguous calls A 17 covers the calls that guidelines don't close: whether this chest pain goes to the ED or home with a stress test next week, whether to start or refuse a controlled substance in a patient with a flagged PDMP, when to override a sepsis alert, and how to work up a vague complaint in someone with six comorbidities — you sign the differential and own the miss, with the shortfall from 20 reflecting how much of primary care runs on USPSTF and specialty algorithms.
The verdict above describes this occupation as a whole. Almost nobody does the typical version of a job — tick what's actually in your week and see how your own mix sits.
Your task mix speaks to task resistance (13/20 here) — how much of the day's work current AI already does. That is the dimension the boxes above are about.
It cannot move the other three. Liability shield (18/20) is whether the law requires a licensed human to sign. Trust premium (16/20) is whether buyers specifically pay for a person. Judgment and accountability (17/20) is whether the role exists to own consequential calls. Those are facts about the occupation's standing, not about which tasks are in your week — a paralegal who does only trial exhibits still holds no licence. Together they are 51 of this occupation's 80 points (64%).
Embodiment (16/20) is also a property of the work rather than the worker, but we don't tag individual tasks as physical or not, so the picker can't tell you anything about it. That's a limit of this tool, not a claim.
Did we get the list right? Tell us what's missing — the tasks are written from the outside, and you're reading this from the inside.
The moves above are yours to make. This is the other half: what would have to change in the world for the occupation itself to score higher. None of it is in any one person's gift, but it is where the floor actually comes from. Scores here are not a one-way ratchet. Only two of the five dimensions — task resistance and embodiment — track what machines can do. The other three track law, what buyers will pay for, and who is answerable, and those move in both directions, often in response to the same pressure AI creates. If every lever below landed, this occupation would score around 87/100, still SAFE.
State boards or DEA rules explicitly requiring that any AI-generated prescription, dosage change, or triage disposition be reviewed and signed by a licensed independent practitioner with personal liability — analogous to the pharmacist-countersign model. Also: full practice authority expansion (now in ~27 states plus VA) makes NPs the accountable signer rather than a physician-supervised extender, moving the shield onto the NP's own license.
Genuine two-tier structure: the clerical tier (notes, prior-auth letters, handouts, coding) is the tier being automated; the residual tier is procedures, exam findings under uncertainty, and dispositions. Panel-size growth per NP raises the judgment density of each remaining hour rather than lowering total task resistance.
If ambient scribes and decision support absorb documentation, coding, and differential generation, the residual day is disposition calls, undifferentiated complaints, and goals-of-care conversations. Watch for malpractice insurers (e.g. NSO/CNA NP policy language) pricing 'AI-assisted' visits with an explicit override-documentation requirement, which formalizes the NP as the accountable decider on every AI recommendation.
Payer or employer contracts guaranteeing a named human clinician per patient panel — e.g. direct primary care and concierge models that market continuity with a specific NP, and CMS primary-care management codes (99490, G2211) that pay for a continuous clinician relationship rather than an encounter.
The limit. Already 80/100; headroom is small and the main downside risk is the mirror of the upside — scope-of-practice liberalization that lets pharmacists, RNs, or AI-supervised protocols handle the routine prescribing tier would erode the licensure shield rather than reinforce it. Employment risk here is panel-size dilution (fewer NPs hired per population), not task substitution.
| New York-Newark-Jersey City, NY-NJ | 21,680 | $163,340 +23% |
| Los Angeles-Long Beach-Anaheim, CA | 10,060 | $161,470 +22% |
| Dallas-Fort Worth-Arlington, TX | 7,310 | $130,980 -1% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 6,590 | $135,700 +3% |
| Chicago-Naperville-Elgin, IL-IN | 6,530 | $131,740 +0% |
| Boston-Cambridge-Newton, MA-NH | 5,990 | $147,480 +11% |
| Houston-Pasadena-The Woodlands, TX | 5,970 | $134,560 +2% |
| Atlanta-Sandy Springs-Roswell, GA | 5,950 | $132,500 +0% |
| San Jose-Sunnyvale-Santa Clara, CA | 1,580 | $231,560 +75% |
| San Francisco-Oakland-Fremont, CA | 3,110 | $219,080 +66% |
| Vallejo, CA | 180 | $182,370 +38% |
Canada Health Infoway · Rush, McLeod Health, Franciscan Missionaries of Our Lady Health System · Northwest Territories Health and Social Services Authority
Canada Health Infoway published results from a national AI scribe program used by primary care clinicians, reporting benefits for documentation work.
Fierce Healthcare reports three health systems — Rush, McLeod Health and FMOL Health — have deployed Suki's AI ambient documentation scribe and report revenue gains from its use.
CBC reports dozens of health-care providers in the Northwest Territories have enrolled in a pilot using AI scribe software for clinical note-taking.
Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.
Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.