← Risk register SOC 21-1023 · reviewed 2026-08-11

Mental Health and Substance Abuse Social Workers

132,810 US workers · median $60,280/yr · Social Service

SAFE

The core of this job — sitting with a person in crisis, conducting biopsychosocial assessments, judging suicide or relapse risk, and holding a therapeutic relationship over months — is exactly what language models cannot own, and much of it happens in homes, hospitals, jails, and shelters rather than at a desk. What is exposed is the paperwork half: progress notes, treatment plan drafting, insurance authorizations, discharge summaries, and resource lookups, which AI already drafts credibly. Clinical practice generally requires LCSW/LMSW licensure with personal liability for the treatment plan, and payers reimburse a named human clinician.

10-year outlook: Demand keeps rising with behavioral health need while AI absorbs the documentation load, so the job gets less clerical and more clinical — but caseloads and pay pressure stay real.

US employment, 2019–2025+12.8%
117,770132,810 workers

Dipped in 2020, then grew past where it started.

Median pay $46,650 → $60,280 +3.4% in real terms (nominal +29.2%, less ~25% US inflation over the period)

The job count is not the verdict

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

+9.7% 136,800 → 150,100 on the projections basis

Hard to automate, and growing

The work resists current AI and the BLS projects +9.7% 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.

~13,500 openings a year on average, including replacing people who leave.

One email if this score changes. Watch as many occupations as you like from the same address — no account, and nothing is sent on a schedule, only when a verdict actually moves.

Also known as — 24 job titles this covers

Titles reported by people doing this work, from the US Department of Labor's O*NET survey. If your job title is here, this page is about your work even though the name doesn't match.

ClinicianCounselorTherapistCase ManagerCrisis WorkerSocial WorkerCase TherapistAlcoholism WorkerDrug Abuse WorkerClinical TherapistAddiction CounselorBehavior SpecialistBehavioral ClinicianGroup Home CounselorMental Health WorkerOutpatient ClinicianOutpatient TherapistAssessment SpecialistBehavioral TechnicianDirect Care CounselorTreatment CoordinatorClinical Social WorkerLicensed Social WorkerBehavioral Case Manager

Score — 70/100 resistance

Holding it up: trust premium (18/20). Weakest point: embodiment (11/20).

Five dimensions, 0–20 each, summed. Higher means more protected. The arithmetic is shown so you can check it: 13 + 11 + 13 + 18 + 15 = 70. · Scored 2026-08-11, and re-examined when evidence accumulates rather than on a schedule.

Task resistance 13/20

Mixed — a routine tier and a judgment tier Roughly half the week is documentation AI already drafts well — DAP notes, ASAM level-of-care justifications, prior-auth letters, 90-day treatment plan updates, referral lookups — while the assessment interview, motivational interviewing with an ambivalent client, and de-escalating someone in withdrawal remain untouched, which is why this sits at 13 rather than 17: the clerical load is genuinely large, not incidental.

Embodiment 11/20

Some physical or field component An 11 reflects that a substantial share of caseload work happens off-site — home visits for outreach and engagement, co-response ride-alongs, jail and ED consults, shelter and detox intakes, transporting or accompanying clients to court — but you are not restraining patients or performing procedures, and telehealth has absorbed a real fraction of individual sessions since 2020.

Liability shield 13/20

Licensed human required and personally liable LCSW/LMSW licensure with state board discipline, mandated-reporter duties under child and elder abuse statutes, and duty-to-warn/Tarasoff exposure put your name personally on the risk assessment and the plan — the reason it is 13 and not 18 is that a meaningful slice of this SOC code works as case managers and bachelor's-level substance abuse counselors under a supervisor's signature rather than independently.

Trust premium 18/20

The human relationship is the product Retention is the intervention here — a client who has been discharged from three programs stays because it is you they call before using, and the working alliance is the single strongest predictor of outcome in the literature, so the relationship is not packaging around a service, it is the service.

Judgment & accountability 15/20

Exists to be accountable for ambiguous calls You decide whether someone gets voluntarily admitted or held involuntarily, whether to break confidentiality under 42 CFR Part 2, whether a parent's relapse triggers a CPS report, and whether a suicidal client is safe to go home tonight — calls made on incomplete information, with days-long consequences, and defensible only by your reasoning.

Confidence: high · reviewed 2026-08-11 · how scoring works

What this job involves — and which parts are yours

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.

AI already does these at usable quality

These still need a person

Active moats on the surviving side: licensure, trust, judgment

How to future-proof this job

Where to go deeper on what this job runs on: Coursera — communication and interpersonal skills free to audit · Coursera — active listening and communication skills free to audit · Toastmasters — public speaking practice at local clubs worldwide low · Khan Academy — reading and vocabulary, all levels, free free · Coursera — critical thinking and logic, audit free free to audit · edX — performance measurement and evaluation free to audit

All 35 skills ranked by how many jobs they open →

What would move this back up — beyond any one person

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 85/100, still SAFE.

6 specific changes that would raise this score
  • already happening liability shield +4

    State boards (e.g., California BBS, Texas BHEC) or ASWB model rules explicitly requiring a licensed LCSW to review, countersign, and take personal responsibility for AI-generated assessments, risk determinations, and treatment plans — mirroring the 2024-25 wave of state AI-in-therapy bills (Illinois' WOPR Act banning AI from independent therapeutic decision-making; Nevada AB 406; Utah HB 452 disclosure rules)

  • already happening task resistance +3

    Task-mix shift: if documentation, prior authorization, and resource lookup are absorbed by AI, the residual caseload concentrates into crisis response, risk formulation, court testimony, and coordination across jails/hospitals/child welfare — the two tiers here are genuinely separable and the judgment tier is the larger one

  • already happening embodiment +2

    Continued or expanded funding for community-based crisis response (988 mobile crisis teams under CMS's Medicaid mobile crisis intervention benefit, CCBHC expansion) pushing more of the work into homes, streets, and shelters rather than telehealth

  • plausible judgment accountability +3

    Statutory duty-to-warn/Tarasoff-style and involuntary-commitment (5150/302) statutes amended to name a licensed human as the only permissible determiner of imminent risk, plus malpractice insurers (NASW Assurance Services) writing exclusions for claims where an AI tool made the risk call

  • plausible liability shield +2

    CMS/Medicaid and commercial payer conditions of participation that make the billed clinician of record personally attest that documentation reflects their own clinical judgment, with AI-drafted notes explicitly non-reimbursable absent attestation — the same mechanism already used for scribe attestation in medicine

  • plausible trust premium +1

    Little headroom: already at 18. Consumer-facing mandatory AI-disclosure rules (Utah HB 452) could make the human-delivered option explicitly marketed, but the premium is largely priced in

The limit. Ceiling is roughly 82-85. Trust premium and embodiment are near saturation; the real movement is liability formalization plus task-mix concentration. Counterpressure to watch: workforce shortage and payer cost pressure pushing AI-assisted or peer-support-plus-AI models into the low-acuity tier, which shrinks headcount even as per-worker resistance rises.

These are conditions, not forecasts — what would have to happen, not what will. Specific rules, cases and bills are named so you can go and check whether they exist and where they stand; verify before relying on any of them. Nothing here is legal or financial advice.

Where this work is, and what it pays there

BLS metro figures for 290 areas. The verdict above does not change by city — the rubric judges what the work involves, not where it happens — but pay and headcount do, and the national median hides a very wide range.

Most of these jobs

New York-Newark-Jersey City, NY-NJ 12,090 $82,450 +37%
Los Angeles-Long Beach-Anaheim, CA 9,380 $78,410 +30%
Boston-Cambridge-Newton, MA-NH 4,660 $63,830 +6%
Phoenix-Mesa-Chandler, AZ 2,870 $46,340 -23%
Dallas-Fort Worth-Arlington, TX 2,720 $47,390 -21%
Minneapolis-St. Paul-Bloomington, MN-WI 2,500 $75,020 +24%
Chicago-Naperville-Elgin, IL-IN 2,450 $57,720 -4%
Washington-Arlington-Alexandria, DC-VA-MD-WV 2,180 $78,210 +30%

Best paid

Vallejo, CA 120 $122,930 +104%
Cheyenne, WY 30 $118,900 +97%
San Jose-Sunnyvale-Santa Clara, CA 1,010 $109,050 +81%

Percentages are against this occupation's national median of $60,280. Counts are jobs in that metro, not vacancies. Metros where the BLS suppressed the cell are absent rather than shown as zero.

Who is actually doing this — nobody, on the record

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.

Read that as a gap in the reporting we can see, not proof of absence — the dispatch runs on English-language feeds and misses plenty. If you know of a case, tell us, or add a field report from inside the job.

Quick take — do you do this job?

Has AI actually changed your work? One tap, anonymous, and the running tally is public. Nothing else is asked of you.

Self-reported and unverified — a sentiment signal, not a survey. One response per person per occupation; you can change your answer.

Field reports — what people say has changed

No field reports yet. A written account takes a paragraph rather than a tap, goes to an editor before it appears, and is the one thing on this page the rubric cannot produce on its own.

File a field report

Concrete beats general: a tool that arrived, a task that moved, a headcount decision you watched happen. Don't include anything that identifies you or your employer if that would put you at risk.

Watch this verdict
Kept current

Rather than check back: get the digest and we'll tell you what changed — or watch a single occupation from its own page.