COOKED
The modal worker here runs a script — phone survey batteries, market research call lists, or hospital admitting intake — and types answers into a fixed form. Conversational AI voice agents and self-service registration kiosks/portals already handle scripted questioning, verification, and structured data capture at usable quality and far lower cost. The surviving sliver is in-person patient registration where an anxious or confused person needs a human at the desk, insurance edge cases need untangling, and someone must physically verify ID and collect signatures.
Part 2020 shock, part continued decline in the years since.
Median pay $34,970 → $45,920 +5.1% 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
-11.6% 164,300 → 145,100 on the projections basis
Exposed, and shrinking
Both signals point the same way: the tasks are largely automatable and the BLS projects -11.6% by 2034. This is the case where the score and the forecast agree, and it is the one worth taking seriously.
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.
~15,800 openings a year on average, including replacing people who leave.
CanvasserEnumeratorCreel ClerkEntry TakerInterviewerCensus ClerkCensus TakerSurvey WorkerData CollectorField ReviewerAdmitting ClerkField CanvasserAdmissions ClerkAdmitting WorkerDesk InterviewerField EnumeratorCensus EnumeratorField InterviewerPolls InterviewerAdmissions AdvisorConsumer RecruiterIntake CoordinatorInterviewing ClerkRegistration Clerk
Holding it up: embodiment . Weakest point: liability shield .
Core tasks are already automatable Reading a CATI script verbatim, coding responses into pre-set categories, dialing from a list, and keying demographics and insurance IDs into a registration form are exactly the tasks voice agents and patient self-check-in portals already do end-to-end — the 4 rather than 0 reflects the in-person admitting desk where you re-explain a form to someone in pain and catch a mismatched policy number before it becomes a denied claim.
Some physical or field component A 6 covers the hospital and clinic registration side — you scan the actual insurance card and photo ID, hand over a wristband, position a signature pad, and walk a patient toward radiology — but the phone-survey and market-research majority of this SOC never leaves a headset and a desk.
No licence, no signature requirement No state licence, no certification exam, no registry: an interviewer needs a high school diploma and employer training, and when a survey response is miscoded or an intake field is wrong, the consequence lands on the research firm's data quality or the hospital's billing office, not on your credential.
Executes defined procedures on defined inputs Your discretion is bounded by the instrument — skip patterns are programmed, probes are scripted, refusal-conversion language is provided, and anything unusual (a patient without ID, a self-pay conversion, a hostile respondent) goes to a supervisor rather than being your call to make.
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 (4/20 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 (1/20) is whether the law requires a licensed human to sign. Trust premium (6/20) is whether buyers specifically pay for a person. Judgment and accountability (3/20) 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 10 of this occupation's 20 points (50%).
Embodiment (6/20) 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.
Credit Counselors EXPOSED
Human Resources Specialists EXPOSED
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 42/100 — EXPOSED.
FCC's February 2024 Declaratory Ruling already classifies AI-generated voices as 'artificial' under the TCPA, requiring prior express consent for outbound calls; if the FCC extends this to cold survey and market-research calls without a research exemption (and states like Florida and Oklahoma add their own AI-caller consent statutes), the outbound telephone-interviewing tier reverts to human dialers by legal necessity rather than capability.
Task-mix shift is genuine here: if kiosks/portals absorb clean scripted registration and clean survey batteries, the residual role is the exception tier — coverage discovery for unclear insurance, Medicare Secondary Payer questionnaires, No Surprises Act good-faith-estimate assembly, and patients who cannot self-serve. Watch hospital revenue-cycle job postings retitling 'registrar' to 'patient access specialist' with financial-counseling duties.
If federal statistical agencies (Census, BLS) formalize a human-administered requirement for flagship surveys — e.g., OMB Statistical Policy Directive guidance barring fully autonomous AI administration of CPS/NHIS instruments on data-quality and nonresponse-bias grounds — the government field-interviewer tier is fenced off from automation.
Identity-proofing rules: if CMS or state Medicaid programs require an in-person human attestation of ID for patient registration (mirroring NIST 800-63 IAL2 supervised-remote/in-person proofing) as an anti-medical-identity-theft control, a named human employee must witness and sign. Some health systems already require this for new-patient registration after fraud audits.
IRB and research-ethics practice: if IRBs routinely require a human interviewer for sensitive-topic protocols (suicidality, intimate partner violence, substance use) because AI cannot execute distress-escalation duties, sponsors pay specifically for a human on those instruments. AAPOR guidance on AI in survey administration is the body to watch.
If patient-access roles formally own charity-care/financial-assistance screening decisions and presumptive-eligibility determinations under IRS 501(r) and state hospital-assistance laws, the registrar makes a consequential, auditable call under ambiguity rather than transcribing one.
If HIPAA enforcement or state consent law treats AI-collected consent-to-treat and financial-responsibility signatures as defective without a human witness — the way notarization rules split on remote online notarization — registration signature capture stays with a human.
The limit. Even with every lever, this occupation is unlikely to clear the 40s. There is no license, no board, and no personal liability to attach to; the shields above protect a shrinking hospital-registration and government-field-interviewer core, not the market-research and CATI phone bank that is the bulk of headcount. The FCC lever protects the calling task from AI while doing nothing to stop those calls being replaced by web panels.
| New York-Newark-Jersey City, NY-NJ | 9,460 | $57,330 +25% |
| Los Angeles-Long Beach-Anaheim, CA | 8,500 | $55,780 +21% |
| Dallas-Fort Worth-Arlington, TX | 4,220 | $47,320 +3% |
| Denver-Aurora-Centennial, CO | 3,100 | $50,180 +9% |
| Houston-Pasadena-The Woodlands, TX | 2,910 | $44,400 -3% |
| San Francisco-Oakland-Fremont, CA | 2,540 | $63,020 +37% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 2,470 | $46,490 +1% |
| Boston-Cambridge-Newton, MA-NH | 2,440 | $50,700 +10% |
| San Jose-Sunnyvale-Santa Clara, CA | 1,110 | $70,840 +54% |
| Vallejo, CA | 120 | $68,830 +50% |
| Barnstable Town, MA | 80 | $66,320 +44% |
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 20. 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.