SAFE
The core product — a sustained therapeutic relationship where a client discloses things they wouldn't type into a chatbot, and a licensed clinician owns the risk assessment — is exactly what AI can't hold. The exposed edges are real: intake questionnaires, progress-note drafting, treatment-plan boilerplate, symptom screening, and scoring of standardized instruments are already largely automatable, and low-acuity supportive counseling faces genuine substitution pressure from AI companion apps. Diagnostic formulation for complex or comorbid presentations, suicide and violence risk calls, and psychological testing interpretation that gets signed and defended remain human.
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
+11.2% 76,300 → 84,800 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +11.2% 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.
~4,800 openings a year on average, including replacing people who leave.
CounselorTherapistCase ManagerPsychologistPsychoanalystElder CounselorGroup CounselorPsychotherapistBehavior AnalystGeropsychologistCriminal ProfilerProgram CounselorTeam PsychologistBehavior TherapistBehavioral AnalystClinical CounselorPersonal CounselorBehavior SpecialistFamily PsychologistHealth PsychologistBehavioral TherapistOutpatient TherapistSchool PsychometristUniversity Counselor
Holding it up: trust premium . Weakest point: embodiment .
Mixed — a routine tier and a judgment tier At 13 rather than 17, the hour-long session and the case formulation behind it resist automation, but the surrounding workload — PHQ-9/GAD-7 administration and scoring, MMPI-2-RF and WAIS scale computation, SOAP-note generation from session audio, insurance treatment-plan templates, and CPT/ICD-10 coding — is already handled by software, and manualized CBT protocols for uncomplicated anxiety or insomnia are the most scriptable part of your caseload.
Some physical or field component A 7 reflects that you work from a chair in an office or over telehealth, with physical presence mattering only in specific contexts — reading agitation and posture during a mental status exam, being in the room during an in-person crisis, administering timed performance subtests with blocks and stimulus books, or working an inpatient unit or forensic facility — none of which involve uncontrolled field conditions.
Licensed human required and personally liable State licensure through a psychology board, a doctoral degree plus supervised hours plus the EPPP, and personal exposure under duty-to-warn statutes (Tarasoff and its state codifications) and mandatory child-abuse reporting law mean your name and licence number sit on the diagnosis, the risk note, and the court-ordered evaluation — a 16 rather than 20 only because bachelor's- and master's-level counselors legally do overlapping talk-therapy work under different credentials.
Exists to be accountable for ambiguous calls A 16 belongs to the person deciding, with incomplete information and no algorithm to hide behind, whether a client is safe to leave the office tonight, whether to initiate an involuntary hold, whether personality pathology or trauma or a medical cause explains the presentation, and whether custody-evaluation findings favor one parent — each defended afterward in a board complaint, deposition, or malpractice suit.
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 (16/20) is whether the law requires a licensed human to sign. Trust premium (19/20) 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 51 of this occupation's 71 points (72%).
Embodiment (7/20) 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 82/100, still SAFE.
State laws restricting AI mental-health services are already in motion — Illinois HB1806 (2025) bars AI from providing therapy or therapeutic decision-making without licensed-professional oversight, Nevada AB406 similar, Utah HB452 imposes disclosure duties. If more states adopt the Illinois model and boards require a licensed psychologist to be the named responsible party for any AI-assisted assessment, formulation, or risk determination, the shield tightens near ceiling.
Genuine two-tier occupation. If intake, screening, scoring, note-drafting and low-acuity supportive contact are absorbed by tools and companion apps, the surviving caseload skews to comorbid, personality-disordered, treatment-resistant, and acute-risk presentations plus supervision of lower-credentialed and AI-assisted providers — the remaining task mix is denser in judgment and measures as more resistant even with no new law. Watch payer carve-outs that route mild anxiety/depression to digital-first care while reserving psychologists for high-acuity tiers.
If AI triage becomes the front door, psychologists increasingly own the escalated calls the model flagged and refused — involuntary hold decisions, duty-to-warn (Tarasoff) determinations, and clinical override of algorithmic risk scores. Formal protocols naming the psychologist as the override authority for AI risk flags, as some health systems already do for sepsis alerts, would push this toward ceiling.
Malpractice carriers (e.g. The Trust, APAIT) adding policy conditions that AI-generated notes, risk screens, or test interpretations must be reviewed and countersigned by the insured clinician, with coverage voided otherwise — a de facto human-signature requirement enforced by insurers rather than boards.
Already near ceiling at 19. Marginal residual route: court and forensic contexts (Daubert/Frye admissibility, custody evaluations, competency and disability determinations) where the retaining party specifically needs a human examiner who can be deposed and cross-examined — if federal or state evidence rules explicitly exclude AI-generated psychological opinion, the forensic segment's premium is locked in.
The limit. trust_premium (19) and liability_shield (16) are already high; realistic headroom is a few points, not a transformation. Embodiment has no plausible upward route — telehealth parity rules have moved this occupation away from physical presence, not toward it. The larger risk to this occupation is volume and price at the low-acuity end rather than displacement of the licensed role itself: if payers and consumers route mild presentations to apps, headcount can fall even as every remaining dimension score rises.
| New York-Newark-Jersey City, NY-NJ | 8,170 | $120,510 +20% |
| Chicago-Naperville-Elgin, IL-IN | 5,030 | $85,390 -15% |
| Los Angeles-Long Beach-Anaheim, CA | 4,790 | $99,250 -1% |
| Boston-Cambridge-Newton, MA-NH | 1,920 | $121,270 +21% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 1,920 | $112,690 +12% |
| San Diego-Chula Vista-Carlsbad, CA | 1,650 | $72,090 -28% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 1,550 | $101,670 +1% |
| San Francisco-Oakland-Fremont, CA | 1,540 | $166,400 +65% |
| San Francisco-Oakland-Fremont, CA | 1,540 | $166,400 +65% |
| San Luis Obispo-Paso Robles, CA | 90 | $155,960 +55% |
| Sacramento-Roseville-Folsom, CA | 630 | $153,050 +52% |
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 71. 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.