COOKED
The daily work — drafting questionnaire items, coding open-ended responses, cleaning datasets, running weighting and crosstabs, writing topline reports and slide decks — is exactly the text-and-tabular pattern work current models handle at usable quality. What resists is sampling-frame design under coverage error, deciding whether a nonresponse-adjusted estimate is defensible enough to publish, and owning the methodology when a client or reporter attacks the numbers. There is no license, no signature requirement, and buyers rarely pay for a relationship with the researcher rather than the finding.
Part 2020 shock, part continued decline in the years since.
Median pay $59,170 → $69,460 -6.1% 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
-5.2% 8,800 → 8,300 on the projections basis
Exposed, and shrinking
Both signals point the same way: the tasks are largely automatable and the BLS projects -5.2% 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.
~700 openings a year on average, including replacing people who leave.
ProctorPollsterResearcherInterviewerData AnalystData CollectorSurvey AnalystField CollectorResearch FellowSurvey AssociateField InterviewerSurvey ResearcherEvaluation AnalystResearch AssociateResearch ScientistSurvey InterviewerResearch SpecialistSurvey StatisticianField Data CollectorField RepresentativeResearch CoordinatorResearch InterviewerSurvey MethodologistRetail Data Collector
Holding it up: judgment & accountability . Weakest point: liability shield .
Mixed — a routine tier and a judgment tier A 7 rather than a 12 because item wording, open-end coding, weighting scripts, and topline write-ups — the bulk of billable hours — are all reproducible by a model, and the genuinely resistant parts (choosing between an address-based frame and a nonprobability panel when coverage is broken, judging whether a 4% response rate estimate can be published at all) are a small slice of the week rather than the spine of it.
Fully desk- and screen-based A 3 rather than 0 acknowledges the occasional intercept survey, cognitive-interview pretest, or field-interviewer training session, but the job is otherwise Qualtrics, Stata/R, and a laptop, with no equipment to handle and no site to visit.
No licence, no signature requirement A 1 reflects that nothing gates entry — no state license, no PSTAT-style credential requirement, no signature block on a deliverable — and AAPOR's Transparency Initiative and Code of Professional Ethics bind by disclosure norms and reputational sanction, not by any statute that could name you personally.
Meaningful discretion A 9 sits in the discretion band because the real calls — post-stratification raking targets, whether to weight a nonprobability sample at all, how to flag straightlining, what caveats go in the methods statement — are yours and are second-guessed by reporters and clients, but they run against established AAPOR and Census guidance and a bad choice yields a criticized poll rather than an injury or a lost case.
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 (7/20 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 (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 16 of this occupation's 26 points (62%).
Embodiment (3/20) 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.
Management Analysts EXPOSED
Business Teachers, Postsecondary 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 41/100 — EXPOSED.
Task-mix shift is genuine here: if item drafting, open-end coding, weighting scripts and topline decks are fully automated, what remains is sampling-frame construction under collapsing response rates, mode-effect adjudication, and detecting AI-generated or bot 'fraudulent respondent' fill in online panels — a problem already forcing panel vendors (Cint, Dynata) to add human-designed fraud detection. That residual is a smaller but harder job.
If polling aggregators, media standards bodies, or litigation-support use force a named human to own the publish/don't-publish call on nonresponse-adjusted estimates — e.g. courts continuing to require a testifying survey expert whose methodology survives Daubert challenge in trademark and class-certification cases, a role that cannot be discharged by an unnamed model output.
If federal statistical policy hardens the OMB Statistical Policy Directive No. 2 / Information Quality Act regime to require a named, credentialed methodologist to attest to the methodology of any survey used in a federal information collection or regulatory record — analogous to how a named responsible official signs OMB clearance packages — a signature requirement would exist where none does. Watch also AAPOR Transparency Initiative disclosure becoming a contractual condition for media publication rather than a voluntary badge.
Narrow route only: if buyers begin paying specifically for attestation that data came from real humans and was analyzed by a named methodologist, as a defense against synthetic-respondent contamination and 'AI-simulated panel' vendors. This is trust in a verifiable warranty, not in a relationship, so the ceiling is low.
The limit. Even with all levers, this stays in the 40s. The occupation is tiny (8,290), has no license to capture, and its core deliverable is a number that clients evaluate on plausibility rather than provenance. Cheap synthetic-respondent substitutes suppress the trust premium from the other side: some buyers will accept simulated panels precisely because they are cheaper, which shrinks the market rather than raising the price of human work.
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 620 | $97,780 +41% |
| Los Angeles-Long Beach-Anaheim, CA | 550 | $83,520 +20% |
| Chicago-Naperville-Elgin, IL-IN | 310 | $54,080 -22% |
| Dallas-Fort Worth-Arlington, TX | 260 | $37,500 -46% |
| Albany-Schenectady-Troy, NY | 210 | $86,890 +25% |
| New York-Newark-Jersey City, NY-NJ | 190 | $87,620 +26% |
| San Francisco-Oakland-Fremont, CA | 180 | $110,540 +59% |
| Boston-Cambridge-Newton, MA-NH | 150 | $74,890 +8% |
| San Francisco-Oakland-Fremont, CA | 180 | $110,540 +59% |
| Ann Arbor, MI | 60 | $103,140 +48% |
| New Haven, CT | 30 | $98,570 +42% |
National University of Singapore
The National University of Singapore announced a strategic collaboration with OpenAI to apply AI across teaching, research and administrative functions.
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.