SAFE
The core of the job is physically guiding a stroke patient through dressing and transfer practice, fitting and fabricating splints, assessing a home or workplace for fall hazards, and adjusting a plan when the person cries or refuses. AI can already draft evaluation narratives, pick CPT codes, generate home exercise handouts, and pre-fill Medicare progress notes — the documentation load that eats OT hours is genuinely exposed. But treatment delivery requires licensed hands in a room, and state licensure plus payer rules make the OT personally accountable for the plan of care.
Dipped in 2020, then grew past where it started.
Median pay $84,950 → $100,330 -5.5% 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
+13.8% 160,000 → 182,100 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +13.8% 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.
~10,200 openings a year on average, including replacing people who leave.
Job TrainerVision TherapistVision SpecialistVocational TrainerMobility SpecialistIndustrial TherapistLow Vision TherapistMobility ProfessionalOrientation SpecialistRehabilitation TeacherRehabilitation EngineerRehabilitation TherapistRehabilitation SpecialistGlobal Mobility SpecialistOccupational Therapist (OT)Assistive Technology TrainerIndependent Living SpecialistCertified Hand Therapist (CHT)Visually Impaired Teacher (TVI)Home Health Occupational TherapistIndustrial Rehabilitation ConsultantVocational Rehabilitation SpecialistCertified Low Vision Therapist (CLVT)Vision Rehabilitation Therapist (VRT)
Holding it up: embodiment . Weakest point: judgment & accountability .
Tasks largely resist digitisation Grading a dressing task in real time, fabricating a thermoplastic wrist orthosis to a specific joint angle, and reading whether a patient's refusal is pain, apraxia, or depression are not steps that survive being turned into a workflow — but goniometry logging, standardized assessment scoring (COPM, FIM), and discharge summary drafting do, which is why this sits at 15 rather than 18.
Hands-on in uncontrolled environments OTs work in ICUs, SNF hallways, school classrooms, and clients' actual kitchens and bathrooms doing home safety assessments — measuring doorway clearances, transferring a 200-pound hemiplegic patient with a gait belt, molding splints on a live limb — environments nobody controls and no fixed equipment covers.
Licensed human required and personally liable State OT licensure under practice acts plus NBCOT certification means you sign the plan of care yourself, your signature is what makes Medicare Part B minutes billable, and a botched transfer or a splint that causes pressure necrosis lands on your licence — not 18+ only because much of the day is delivered under a physician referral and an OTA can execute treatment under your supervision.
Exists to be accountable for ambiguous calls You decide when someone is safe to go home alone, whether cognition supports independent medication management, and when to discharge or escalate against family pressure — high-stakes and genuinely ambiguous, though bounded by physician orders, established frames of reference, and payer-driven visit caps.
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 (15/20 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 (15/20) 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 (15/20) 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 47 of this occupation's 79 points (59%).
Embodiment (17/20) 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 92/100, still SAFE.
Task-mix shift: once documentation, CPT coding, HEP generation and Medicare progress-note pre-fill are fully automated (already in motion via Netsmart/WebPT/Raintree AI scribes), the residual job is almost entirely hands-on assessment, splint fabrication, home-hazard judgment and behavioral de-escalation — the tier AI cannot reach. This raises the share of the day that is AI-resistant without any new law.
Growth of cash-pay niches where the buyer is explicitly purchasing a human's hands and presence — pediatric sensory/feeding therapy, hand therapy (CHT), lymphedema, and private home-safety consulting for aging-in-place. Parents and adult children of aging clients already reject app-delivered substitutes; if payer coverage stays thin these segments grow as out-of-pocket markets.
CMS tightening the Medicare Part B requirement that an OT (not an OTA or AI-assisted workflow) personally perform and sign the initial evaluation, reassessment, and discharge — plus state practice acts adding explicit language that AI-generated plans of care require licensed OT authorship and personal attestation. AOTA has already pushed model language on AI documentation attestation; several state boards (e.g., telehealth-era rule updates) are the venue to watch.
Expansion of OT sign-off authority on consequential calls: driver-fitness and return-to-work determinations, guardianship/capacity input, and discharge-destination safety recommendations. State DMV medical advisory board rules that name OT driving rehab specialists as the assessor of record, and workers' comp statutes accepting OT functional capacity evaluations as binding, both already exist in some states and could widen.
Malpractice carriers and hospital credentialing bodies adding a condition that AI-drafted evaluations/discharge recommendations are only covered if countersigned by the treating OT with documented independent review — the same mechanism radiology insurers have used for autonomous read tools.
The limit. Already at 79 with high embodiment and trust; realistic headroom is a few points at most. The countervailing pressure is real: if AI documentation makes each OT able to carry a larger caseload, or if payers push more delivery to OTAs and aides supervised remotely with AI-generated plans, headcount can fall even as the remaining role scores higher on every dimension. A high score protects the occupation's character, not its worker count.
| New York-Newark-Jersey City, NY-NJ | 11,640 | $101,460 +1% |
| Chicago-Naperville-Elgin, IL-IN | 5,960 | $99,860 +0% |
| Los Angeles-Long Beach-Anaheim, CA | 4,950 | $122,040 +22% |
| Boston-Cambridge-Newton, MA-NH | 4,790 | $101,940 +2% |
| Dallas-Fort Worth-Arlington, TX | 3,450 | $106,110 +6% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 3,410 | $103,140 +3% |
| Houston-Pasadena-The Woodlands, TX | 2,700 | $103,580 +3% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 2,640 | $102,960 +3% |
| San Jose-Sunnyvale-Santa Clara, CA | 840 | $136,870 +36% |
| San Francisco-Oakland-Fremont, CA | 1,560 | $136,770 +36% |
| Vallejo, CA | 100 | $135,000 +35% |
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 79. 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.