SAFE
The documentation half of this job — case notes, service plans, eligibility paperwork, referral letters, court report drafts — is exactly what language models do well, and caseload software is already absorbing it. The other half is home visits, in-person child safety assessments, sitting with a family in crisis, testifying in dependency court, and deciding whether a child stays in the home; none of that transfers to software, and state licensure plus mandated-reporter liability keeps a named human on the file. Modal worker here is a public-agency or school-based caseworker, not a private clinician, so the licensure moat is real but uneven across states.
Headcount grew steadily across the period.
Median pay $47,390 → $59,550 +0.5% 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
+3.4% 399,900 → 413,300 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +3.4% 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.
~35,100 openings a year on average, including replacing people who leave.
CaseworkerCase WorkerCase ManagerGroup WorkerSocial WorkerAdoption AgentChild AdvocateFamily ManagerYouth AdvocateAdoption WorkerFamily AdvocateInterventionistParent EducatorJuvenile OfficerSchool TherapistYouth SpecialistFamily CaseworkerAdoption CounselorChild Abuse WorkerFoster Care WorkerJuvenile CounselorAdoption SpecialistCasework SupervisorJuvenile Specialist
Holding it up: trust premium . Weakest point: liability shield .
Mixed — a routine tier and a judgment tier At 13 rather than 16, the ceiling is set by how much of the week is genuinely text — CANS/SDM risk instruments, ICWA notice paperwork, Title IV-E eligibility redeterminations, monthly contact logs, IEP-related documentation — while the irreplaceable core is the unannounced home visit, the kinship-placement conversation at 9pm, and the dependency-court testimony that no model can be sworn in to give.
Hands-on in uncontrolled environments You drive to trailer parks and third-floor walkups you've never seen, look in the refrigerator and at the sleeping arrangements, smell the apartment, physically transport a child to a receiving home, and get between arguing adults — 13 reflects that the assessment is sensory and unscheduled, though you're not lifting patients or operating equipment, which is what keeps it out of the high teens.
Licensed human required and personally liable 11 sits at the bottom of the licensed band because it is real but jagged: LMSW/LCSW is required for clinical and many school positions and the board can sanction your license, and mandated-reporter statutes attach criminal exposure to your name personally, but plenty of states still hire county child-welfare caseworkers with a bachelor's and a training academy certificate, so the credential doesn't gate the whole occupation.
Exists to be accountable for ambiguous calls 16 is right because you make the removal-or-leave call on incomplete information, often within hours and often alone in the driveway, and the fatality reviews, news coverage, and civil suits that follow a wrong call in either direction name the caseworker's decision — SDM tools structure that call but do not own it.
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 (11/20) 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 (16/20) 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 44 of this occupation's 70 points (63%).
Embodiment (13/20) 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.
Genuine two-tier structure: if caseload software absorbs notes, eligibility forms and referral letters, the residual job concentrates on home visits, crisis de-escalation, court testimony and removal decisions — the score rises by task-mix shift with no new law, though usually alongside caseload increases per worker.
Juvenile/dependency court rules requiring that court reports and testimony be attested by the named worker who personally conducted the home visit, with a certification that no unverified automated content was included — analogous to federal Rule 11 AI-certification standing orders already issued by dozens of US district judges.
State child-welfare statutes or CFSR/ACF program instructions requiring that any AI-generated risk score or safety-decision recommendation be countersigned by a licensed (LCSW/LMSW) caseworker who retains personal mandated-reporter liability — the pattern already visible in Allegheny County AFST governance rules and in state bills (e.g. California AB 331-style algorithmic accountability proposals) requiring human review of consequential automated decisions. Also: states that still allow bachelor's-level unlicensed caseworkers moving to mandatory licensure for anyone signing a removal recommendation.
Post-fatality reform legislation (the recurring pattern after high-profile child deaths) that names an individual worker of record accountable for each safety decision and bars delegation of the removal call to a screening algorithm; conversely, formal 'decision of record' designations in state practice manuals.
Union contracts (AFSCME, SEIU locals representing county child-welfare staff) bargaining minimum in-person contact standards and caps on AI-substituted client contact; school districts contractually guaranteeing families a named human social worker rather than a triage chatbot.
The limit. trust_premium is already 17 — the modal worker is public-agency, so 'buyers' are agencies under budget pressure, not families who can pay for a human; there is little headroom. Embodiment is intrinsically capped near current level since the physical component is presence, not manipulation.
| Los Angeles-Long Beach-Anaheim, CA | 26,220 | $71,250 +20% |
| New York-Newark-Jersey City, NY-NJ | 24,350 | $66,540 +12% |
| Chicago-Naperville-Elgin, IL-IN | 13,550 | $64,790 +9% |
| Boston-Cambridge-Newton, MA-NH | 7,070 | $77,420 +30% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 6,850 | $55,930 -6% |
| Detroit-Warren-Dearborn, MI | 6,540 | $59,270 +0% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 6,490 | $75,550 +27% |
| San Francisco-Oakland-Fremont, CA | 5,840 | $72,020 +21% |
| San Jose-Sunnyvale-Santa Clara, CA | 2,120 | $87,600 +47% |
| Norwich-New London-Willimantic, CT | 450 | $81,830 +37% |
| Salisbury, MD | 120 | $81,520 +37% |
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.