SAFE
Bathing, transferring, toileting, dressing, meal prep, and wound-site observation in a cluttered private home are exactly the tasks robotics cannot touch and language models cannot substitute for. The automatable slice is thin — visit notes, scheduling, care-plan documentation, and eligibility paperwork — while the hands-on hours are the job. The real threats here are wages and turnover, not AI; note that licensure is weak (agency training and state HHA certification rather than a personally liable license), so the moat is physical and relational, not regulatory.
Headcount grew steadily across the period.
Median pay $25,280 → $35,800 +13.3% 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
+17% 4,347,700 → 5,087,500 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +17% 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.
~765,800 openings a year on average, including replacing people who leave.
AideSitterCaregiverCaretakerCompanionBlind AideCare WorkerBlind EscortHospice AideCare ProviderHome ProviderCare CaregiverCare CompanionGeriatric AideHome AttendantHome Care AideHome CaregiverHome Help AideNutrition AidePatient SitterHome Care GiverMedication AideResidential AideElderly Companion
Holding it up: embodiment . Weakest point: liability shield .
Tasks largely resist digitisation Pivot-transferring a 200-pound contracture patient from bed to commode, coaxing a dementia client through a shower, and pressure-cooking a low-sodium meal in someone else's kitchen are the hours that get billed — the software-shaped slice is the EVV clock-in and the daily task checklist, which is why this sits at 17 rather than mid-teens.
Hands-on in uncontrolled environments Every shift happens in an uncontrolled private residence with unpredictable stairs, pets, hoarding, and no lift equipment, and the work is skin-to-skin: perineal care, nail and mouth care, repositioning, ambulation support — there is no remote version of any of it, which is the definition of 20.
Certification preferred, not legally required Medicare-certified agencies require 75 hours of training and a competency evaluation under 42 CFR 484.80, and some states add an HHA registry listing, but the aide holds no license that can be revoked by a board and the RN who signs the care plan owns clinical liability — that training floor is what lifts this off 3 to a 7.
Meaningful discretion The care plan specifies the ADLs and the aide is barred from adjusting medication or performing skilled tasks, but the discretion that matters is unsupervised triage — noticing a new sacral redness, a sudden confusion, a fall risk, or a bruise that means calling the nurse or APS — real stakes, narrow authority, which is a 9 rather than a 14.
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 (17/20 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 (7/20) 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 (9/20) 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 33 of this occupation's 70 points (47%).
Embodiment (20/20) 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 84/100, still SAFE.
Electronic Visit Verification (21st Century Cures Act, already in force for Medicaid personal care) tightened so the aide's own attestation of task completion becomes the fraud-liable record, making the named individual — not the agency — the signer on billable service delivery
Consumer-directed care budgets (Medicaid self-direction, VA Veteran-Directed Care) growing so the client personally hires and can refuse remote-monitoring substitution — buyer choice of a specific named human is contractual, not incidental
If sensor/remote-monitoring platforms absorb the documentation, scheduling and eligibility paperwork slice, the residual day is entirely hands-on and relational — task-mix shift with no law needed
State expansion of mandatory HHA certification with individual registry listing and personal accountability for abuse/neglect findings — e.g. states adopting CMS-style training-hour minimums for Medicaid HCBS personal care workers, or California-style Home Care Aide Registry requirements extended to all payers, plus mandated-reporter status with individual sanction rather than agency-only liability
Hospital-at-home and Medicaid HCBS programs formally assigning aides escalation duties — protocolized deterioration triage, fall-risk and skin-integrity reporting that nurses act on — as in some ACO/managed-LTSS contracts paying differentials for 'care escalation' roles; the aide then owns the call to escalate under ambiguity
The limit. Embodiment is already maxed and task_resistance is near ceiling; realistic headroom is almost entirely in liability_shield and judgment_accountability. None of these levers touch the actual threat to this occupation, which is wage suppression, immigration restriction on the labor supply, and 60%+ annual turnover — a higher AI-resistance score here does not mean a better job.
| New York-Newark-Jersey City, NY-NJ | 644,700 | $39,680 +11% |
| Los Angeles-Long Beach-Anaheim, CA | 385,530 | $34,320 -4% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 158,280 | $29,800 -17% |
| San Francisco-Oakland-Fremont, CA | 119,120 | $35,950 +0% |
| Chicago-Naperville-Elgin, IL-IN | 108,520 | $38,180 +7% |
| Riverside-San Bernardino-Ontario, CA | 102,760 | $34,320 -4% |
| Minneapolis-St. Paul-Bloomington, MN-WI | 93,560 | $38,860 +9% |
| Boston-Cambridge-Newton, MA-NH | 72,420 | $40,910 +14% |
| Grants Pass, OR | 4,640 | $53,120 +48% |
| Seattle-Tacoma-Bellevue, WA | 54,670 | $48,210 +35% |
| Longview-Kelso, WA | 1,780 | $47,840 +34% |
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 70. 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.