SAFE
Neurology's diagnostic core rests on the hands-on neurological exam — reflexes, gait, tone, cranial nerves, subtle asymmetries — which no current system can perform, plus procedures like EMG/nerve conduction, lumbar puncture, botulinum injections, and EEG supervision. AI is genuinely strong on the pattern layer: MRI lesion detection, EEG spike flagging, differential generation, and note drafting, so documentation and first-pass image reads will compress. What persists is owning ambiguous calls — is this MS or an atypical mimic, do we thrombolyse, do we escalate immunotherapy in a deteriorating encephalitis patient — under personal license and liability, and delivering diagnoses like ALS or dementia to families over years of care.
Headcount grew steadily across the period.
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.
BLS projection, 2024–2034
+5.4%
Percentage only. The projection counts a different population from the 10,590 above — it includes self-employed workers, which for this occupation is most of them, so the two headcounts are not comparable.
Hard to automate, and growing
The work resists current AI and the BLS projects +5.4% 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.
~300 openings a year on average, including replacing people who leave.
PhysicianNeurologistEpileptologistNeurohospitalistAdult NeurologistChild NeurologistNeurophysiologistGeneral NeurologistHeadache SpecialistMD (Medical Doctor)Neurology PhysicianVascular NeurologistPediatric NeurologistNeurosurgery PhysicianOsteopathic NeurologistChiropractic NeurologistPediatric Neurology PhysicianAdult and Pediatric NeurologistDO Physician (Doctor of Osteopathic Medicine Physician)
Holding it up: liability shield . Weakest point: task resistance .
Mixed — a routine tier and a judgment tier A 12 reflects that a real share of the neurologist's day — dictating H&Ps, generating differentials from a symptom list, first-pass reads of MRI for white-matter lesions, screening 24-hour EEG for spikes, titrating antiepileptics by protocol — is already being absorbed by ambient scribes and FDA-cleared triage software, while localizing a lesion from an inconsistent exam or distinguishing psychogenic nonepileptic seizures from frontal-lobe events at the bedside stays out of reach.
Hands-on in uncontrolled environments A 14 rather than 18 recognizes that the hands are essential but the environment is mostly controlled: percussing reflexes, testing tone and pronator drift, needle EMG through muscle, LP into the L3-L4 space, EMG-guided botulinum into cervical dystonia — all in clinic or on the ward, not roadside or in a patient's kitchen, and a meaningful fraction of stroke-call decisions now happen over telestroke video.
Licensed human required and personally liable A 19 is set by the fact that tPA and thrombectomy decisions, DEA-scheduled prescribing of antiepileptics and stimulants, and EMG interpretation are signed under an individual state medical license backed by board certification in neurology, and a door-to-needle miss or a missed cord compression names the neurologist personally in the malpractice complaint, not the hospital's algorithm vendor.
Exists to be accountable for ambiguous calls An 18 tracks the calls with no clean rule: thrombolysis in a 79-year-old on apixaban with an unclear last-known-well, immunosuppression escalation in seronegative autoimmune encephalitis that is deteriorating, brain-death determination and apnea testing, and whether an atypical MRI is MS, NMOSD, or sarcoid — each made on incomplete data with irreversible consequences.
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 (12/20 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 (19/20) is whether the law requires a licensed human to sign. Trust premium (17/20) is whether buyers specifically pay for a person. Judgment and accountability (18/20) 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 54 of this occupation's 80 points (68%).
Embodiment (14/20) 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 89/100, still SAFE.
As AI absorbs first-pass MRI reads, EEG spike flagging, and note drafting, the residual day concentrates on the ambiguous tier: atypical demyelinating vs mimic, immunotherapy escalation in autoimmune encephalitis, thrombolysis/thrombectomy calls under time pressure, and multi-year disease-modifying therapy sequencing. This is a genuine two-tier occupation and the judgment tier is not currently automatable.
Payer and health-system telestroke contracts that specify a board-certified vascular neurologist must be on the video call for thrombolysis decisions, plus Joint Commission Comprehensive Stroke Center certification standards naming neurologist availability, convert trust into a purchased requirement rather than a preference.
Growth in procedure-anchored subspecialty work — EMG/NCS, botulinum chemodenervation for dystonia/spasticity, DBS programming, intrathecal nusinersen, epilepsy surgery mapping — shifts the case mix toward tasks requiring hands and real-time interpretation. Watch RVU mix and fellowship match data in neuromuscular and epilepsy.
FDA moving toward requiring autonomous-mode neuro imaging/EEG triage devices (e.g. Viz.ai LVO, Persyst) to be labeled adjunctive-only with a named physician-of-record interpretation, plus CMS conditioning the AI add-on payment (NTAP-style) on documented physician concurrence, would harden the countersignature requirement.
State board and malpractice-carrier positions that AI output does not transfer the standard of care — mirroring recent AMA policy on 'augmented intelligence' physician oversight — plus court rulings holding the interpreting neurologist liable for a missed AI-flagged lesion, keep the consequential call personally owned.
The limit. Already 80/100; the institutional dimensions are near saturation, so realistic headroom is a few points, mostly from task-mix concentration in the judgment and procedural tier. The downward pressure on routine imaging/EEG reads is real and may offset.
| Dallas-Fort Worth-Arlington, TX | 300 | $360,640 +45% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 190 | $161,560 -35% |
| Cincinnati, OH-KY-IN | 180 | $101,210 -59% |
| Chicago-Naperville-Elgin, IL-IN | 170 | $136,030 -45% |
| Seattle-Tacoma-Bellevue, WA | 130 | $331,080 +33% |
| Houston-Pasadena-The Woodlands, TX | 120 | $277,960 +12% |
| Minneapolis-St. Paul-Bloomington, MN-WI | 120 | $335,420 +35% |
| Charlotte-Concord-Gastonia, NC-SC | 90 | $337,600 +36% |
| Memphis, TN-MS-AR | 30 | $535,950 +116% |
| Buffalo-Cheektowaga, NY | 80 | $451,510 +82% |
| Phoenix-Mesa-Chandler, AZ | 70 | $409,610 +65% |
We have no reported case of a named organisation automating this occupation. Not one deployment, not one announcement.
That is worth saying out loud next to a score of 80. The verdict above is about what the work exposes — what current AI could do to these tasks. It is not a claim that anyone has done it. For this occupation those two things have come apart completely: the capability argument is on this page, and the evidence column is empty.
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.