COOKED
The core output — literature reviews, policy memos, survey instrument drafting, comparative country analyses, legislative and regulatory tracking — is text-in/text-out work that current models produce at usable draft quality, and much of the quantitative work is standard regression and coding of public datasets. No license gates the title, and clients rarely pay for the specific human. What survives is the senior advisory tier: someone who has cultivated sources, reads a political situation that isn't in the training data, and puts their name on a forecast a principal will act on.
This fall is concentrated in 2020 and has not recovered since.
Median pay $122,220 → $142,080 -7.0% 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
-3.1%
Percentage only. The projection counts a different population from the 5,540 above — it includes self-employed workers, which for this occupation is most of them, so the two headcounts are not comparable.
Exposed, and shrinking
Both signals point the same way: the tasks are largely automatable and the BLS projects -3.1% 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.
~500 openings a year on average, including replacing people who leave.
StrategistPolicy AdvisorPolicy AnalystPolicy OfficerPolitical AidePolicy AssociatePolicy SpecialistPolitical AdvisorPolitical AnalystLegislative AnalystLegislative LiaisonPolitical ConsultantPolitical ResearcherHealth Policy AnalystMedical Policy AnalystProject Policy AnalystLegislative Policy AnalystLocal Governance SpecialistPolitical Research ScientistGovernment Affairs ResearcherGovernment Affairs SpecialistLegislative Affairs SpecialistCitizen Participation Specialist
Holding it up: judgment & accountability . Weakest point: liability shield .
Mixed — a routine tier and a judgment tier An 8 reflects that coding open-ended survey responses, drafting country risk briefs, tracking bills through committee, and running cross-national regressions on Polity or V-Dem data are all now first-draftable by a model, while what stays human is elite interviewing, fieldwork in a country where the officials won't talk to a stranger, and designing a research question nobody has framed yet — enough of the job to keep it out of the 0-6 band but not enough to reach mixed-territory 12.
Fully desk- and screen-based A 4 covers the occasional overseas fieldwork trip, focus group facilitation, or archival visit, but the modal week is Stata, Qualtrics, a literature database, and a Word document, and no part of the analysis requires the researcher's body to be anywhere in particular.
No licence, no signature requirement A 1, not a 0, because nothing licenses the title — a PhD is a hiring credential, not a statutory gate — and when a forecast is wrong the institution absorbs it; the only faint personal exposure is IRB approval on human-subjects protocols and security clearance obligations for government contract work.
Meaningful discretion An 11 marks the gap between the methodologically constrained work — model specification, sampling frames, coding rules that peer review polices — and the genuinely unbounded calls like whether a coup is likely in the next six months or whether a sanctions regime will hold, where the analyst commits to a judgment that a principal acts on but where the decision itself, and its consequences, belong to someone else.
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 (8/20 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 (9/20) is whether buyers specifically pay for a person. Judgment and accountability (11/20) 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 21 of this occupation's 33 points (64%).
Embodiment (4/20) 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.
Law 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 46/100 — EXPOSED.
Task-mix shift: as literature reviews, regulatory tracking and standard regression are absorbed by models, the remaining paid work concentrates in the elicitation tier — expert interviews with in-country sources, closed-door elite briefings, and interpretation of events with no textual precedent (coup, contested election, wartime negotiation). Watch for whether intelligence-community and geopolitical-risk shops (Eurasia Group, Rand, Foreign Service INR) redefine the analyst job description around primary-source elicitation rather than synthesis.
Named-analyst attribution requirements in the products buyers actually pay for: if institutional investors and asset managers demand a signed human analyst-of-record on political-risk notes the way sell-side equity research must carry an analyst name under FINRA 2241 / MiFID II research-unbundling rules, the buyer is explicitly paying for the person. Also watch whether journals adopt strict authorship rules barring generative-AI-drafted analysis, and whether expert-witness and congressional-testimony demand (which requires a live human under oath) grows as a share of billings.
Formalized forecast accountability: if government and philanthropic funders require scored, attributable probabilistic forecasts from a named principal investigator — the model of IARPA's ACE/HFC tournaments and ODNI analytic-standards ICD 203 traceability requirements extended to contracted outside analysts — the role owns a consequential call under ambiguity with a track record attached to a person.
Only a narrow route: FARA registration and lobbying-disclosure regimes require a named human filer, and if expert-witness admissibility rules (post-2023 FRE 702 amendment) are applied to bar AI-generated political analysis without a qualified human expert who can be cross-examined, testimony work becomes human-gated. This does not touch the memo-writing bulk of the occupation.
The limit. Realistic ceiling is roughly the mid-40s. There is no licensure body for the title and none is being proposed, so liability_shield cannot move far; embodiment has no route at all. Any gain is concentration into a small senior advisory tier, which means the score can rise while headcount falls.
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 3,700 | $156,730 +10% |
| Boston-Cambridge-Newton, MA-NH | 190 | $133,930 -6% |
| New York-Newark-Jersey City, NY-NJ | 110 | — |
| Chicago-Naperville-Elgin, IL-IN | 90 | $103,780 -27% |
| Seattle-Tacoma-Bellevue, WA | 80 | $153,960 +8% |
| Ann Arbor, MI | 60 | $86,290 -39% |
| Phoenix-Mesa-Chandler, AZ | 30 | $114,320 -20% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 3,700 | $156,730 +10% |
| Seattle-Tacoma-Bellevue, WA | 80 | $153,960 +8% |
| Boston-Cambridge-Newton, MA-NH | 190 | $133,930 -6% |
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 33. 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.