COOKED
The core of this job — keying application data, verifying policy fields against forms, calculating premiums from rate tables, routing claims to adjusters, and issuing correspondence letters — is exactly the structured document-and-form work that OCR plus current language models already handle at production quality inside carrier core systems. There is no licensure requirement, no physical presence, and virtually no discretionary authority: coverage decisions and claim payouts belong to licensed adjusters and underwriters, not to processing clerks. The thin surviving tier is exception handling and phone contact with confused or upset policyholders, and that tier is much smaller than current headcount.
This fall is concentrated in 2020 and has not recovered since.
Median pay $40,750 → $49,230 -3.4% 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
-3.7%
Percentage only. The projection counts a different population from the 214,260 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 -3.7% 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.
~20,300 openings a year on average, including replacing people who leave.
ReviewerClaims ClerkClaims TakerClaims SorterRevival ClerkClaims AnalystPolicy AnalystPolicy CheckerInsurance ClerkClaims AssistantClaims AssociateClaims ProcessorInsurance BillerProcessing ClerkDocument ReviewerInsurance AnalystInsurance CheckerReinsurance ClerkCancellation ClerkClaims AdjudicatorClaims CoordinatorDocument ProcessorPolicy Issue ClerkUnderwriting Clerk
Holding it up: judgment & accountability . Weakest point: liability shield .
Core tasks are already automatable Transcribing ACORD forms into Guidewire or Duck Creek, cross-checking VINs and effective dates against declarations pages, running rate-table premium math, and generating cancellation and reinstatement notices from templates are all deterministic field-to-field operations that carriers already run through OCR-plus-LLM straight-through processing pipelines — the 3 rather than 8 reflects that even the messy inputs (faxed forms, handwritten supplements) are now readable by machine, leaving no task tier that structurally requires a person.
Fully desk- and screen-based The work is a dual-monitor workstation, a headset, and a scanner queue — sometimes a physical mail run or a file room pull at older carriers, which is why this is 2 and not 0, but nothing about the job puts you in a place a camera and an API cannot reach.
No licence, no signature requirement No state requires a licence to key a policy application or index a first-notice-of-loss; the adjuster licence and the resident producer licence sit with other people in the building, and if you mis-key a coverage limit the carrier's E&O and the licensed underwriter's file review absorb it — the 2 rather than 0 is only for carriers that require an AINS or similar internal designation as a pay-grade condition, which no statute backs.
Executes defined procedures on defined inputs You decide which adjuster queue a loss belongs in, whether a proof-of-loss packet is complete enough to advance, and when to escalate a suspected duplicate claim — real calls, but bounded by claim-handling manuals and unfair-claims-practice timelines, and the coverage determination, reserve setting, and denial letter are signed by someone else, which puts this at 4 rather than 9.
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 (3/20 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 (2/20) is whether the law requires a licensed human to sign. Trust premium (3/20) is whether buyers specifically pay for a person. Judgment and accountability (4/20) 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 9 of this occupation's 14 points (64%).
Embodiment (2/20) 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 27/100, still COOKED.
Task-mix shift: if straight-through processing absorbs keying, rate-table math and correspondence, the residual role becomes exception triage — fraud-flag reconciliation, mismatched-document disputes, and AI-output QA (reviewing model-generated denials against policy language). The job genuinely has this second tier, but it supports far fewer seats; score rises per surviving worker while headcount falls.
If carriers formalize an 'AI reviewer of record' role in claims workflow — a person logged in the file as having overridden or approved a model recommendation, discoverable in bad-faith litigation — the surviving clerk seats own consequential calls rather than routing them. Watch for insurer E&O carriers or state market-conduct exams requiring named-reviewer audit trails.
State adoption of NAIC-style AI model bulletins (already issued in ~20+ states) hardening into rules that require a named human reviewer to attest to any adverse or automated claim/policy action — plus Colorado SB21-169 quantitative-testing regime and California SB 1120-style 'no algorithm-only denial' laws extending from health to P&C. This only helps clerks if the attesting human is the processing staffer rather than a licensed adjuster/underwriter, which is the weaker half of the bet.
Some route exists in complaint/appeal channels: state DOI complaint-handling and NAIC market conduct standards effectively require a reachable human on escalated files, and union/contract language at carriers like the UAW-represented MetLife/Farmers units has protected live-contact staffing. This is thin — buyers pay for the policy, not for the clerk.
The limit. Even with every lever, this tops out in the low-to-mid 30s and the realistic ceiling is on the per-worker score, not on headcount: the licensure and liability that exist in insurance attach to adjusters and underwriters, and any new human-attestation rule is far likelier to be satisfied by upgrading those licensed roles than by preserving unlicensed clerical seats.
| New York-Newark-Jersey City, NY-NJ | 7,700 | $60,070 +22% |
| Los Angeles-Long Beach-Anaheim, CA | 7,500 | $50,630 +3% |
| Dallas-Fort Worth-Arlington, TX | 7,100 | $57,340 +16% |
| Phoenix-Mesa-Chandler, AZ | 6,770 | $50,620 +3% |
| Chicago-Naperville-Elgin, IL-IN | 5,460 | $51,280 +4% |
| Austin-Round Rock-San Marcos, TX | 3,450 | $48,280 -2% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 3,450 | $57,150 +16% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 3,350 | $47,520 -3% |
| Manchester-Nashua, NH | 270 | $67,170 +36% |
| San Francisco-Oakland-Fremont, CA | 2,390 | $64,260 +31% |
| Barnstable Town, MA | 70 | $63,190 +28% |
Chubb · Allianz · Ergo · Suncorp
Yahoo Finance reports that insurer Allianz plans job cuts linked to AI automation of insurance roles.
Crypto Briefing reports that insurer Allianz plans to cut 1,800 jobs, attributing the reduction to AI automation in insurance operations.
Insurance Business reported that Chubb's CEO signalled significant workforce reductions tied to the insurer's accelerating AI strategy.
Insurance Business reported that Munich Re subsidiary Ergo announced an AI-driven workforce restructuring plan affecting around 1,000 positions by 2030.
Insurance Business reports that insurer Chubb plans to cut up to 20% of its workforce as part of an AI-driven restructuring.
Insurance Business reported that Australian insurer Suncorp is giving AI a larger role in its business strategy.
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.