COOKED
The core of this job — keying data, filing and retrieving records, drafting routine correspondence, processing invoices and forms, answering and routing calls — is exactly the text-and-screen work current AI plus workflow software already handles at usable quality. What survives is physical: sorting and distributing mail, handling paper records and equipment, stocking supplies, walking documents between people, and being the person at the front desk when someone shows up. There is no license, no signature requirement, and almost no ambiguous-decision authority to anchor the role.
This fall is concentrated in 2020 and has not recovered since.
Median pay $34,040 → $45,010 +5.8% 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
-6.7% 2,646,000 → 2,468,200 on the projections basis
Exposed, and shrinking
Both signals point the same way: the tasks are largely automatable and the BLS projects -6.7% 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.
~282,400 openings a year on average, including replacing people who leave.
ClerkRouterCharterStubberMap ClerkDesk ClerkUnit ClerkYard ClerkChart ClerkField ClerkFloor ClerkMedia ClerkOffice AidePolice AideTrace ClerkTrain ClerkMelter ClerkOffice ClerkOrder CallerPolice ClerkClerical AideGeneral ClerkGrading ClerkLaundry Clerk
Holding it up: embodiment . Weakest point: liability shield .
Core tasks are already automatable At 5, the bulk of the day — keying entries into the AP or student-records system, pulling and re-filing folders, typing form letters from templates, reconciling a batch of invoices against POs, transferring calls — is already done end-to-end by OCR intake, RPA scripts and auto-attendants in shops that have bought them; it sits above the 0-4 floor only because mail runs, jammed copiers, walk-in visitors and the ad-hoc 'can you find that 2019 file in the basement' requests still need a body.
Some physical or field component A 7 reflects that the physical work is real but indoors and low-skill: sorting and distributing interoffice mail, carrying boxes of records to storage, loading the postage meter, restocking the supply cabinet, staffing a front counter — controlled climate, no ladders, no field sites, nothing that would push it toward the 13+ band of technicians working in someone else's building.
No licence, no signature requirement A 1 is honest: no state licenses general office clerks, nothing you produce carries your signature as the responsible party, and when a misfiled record or a mistyped invoice amount surfaces it is the office manager, controller or attorney of record who answers for it — the only reason it isn't 0 is that clerks handling PHI or student records are personally bound by HIPAA and FERPA confidentiality rules.
Executes defined procedures on defined inputs A 3 matches work governed by the desk manual: which retention schedule a document falls under, which department a call goes to, which coding a routine invoice gets — all written down somewhere, with anything unusual escalated upward rather than decided, and the discretion that exists is prioritizing your own queue when three people need something at once.
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 (5/20 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 (4/20) 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 8 of this occupation's 20 points (40%).
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 32/100, still COOKED.
The residual physical tier — mail sortation, paper records custody, badge/visitor handling, supply stocking, walking wet-signature documents between offices — becomes the whole job as screen tasks are absorbed. Watch for job postings retitled 'office services associate' or 'facilities/mailroom coordinator' with no data-entry duties listed. Sectors still holding paper originals (courts, county recorders, medical records under state retention statutes, notarized real-estate files) sustain this longest.
Two-tier split: routine keying automates, and what remains is exception handling — reconciling records that don't match, chasing a vendor whose invoice fails three-way match, dealing with the walk-in whose paperwork is wrong. This tier is genuinely resistant because it is where the system already failed. Recognizable if headcount falls sharply while remaining roles get 'specialist' or 'coordinator' titles and higher pay bands.
Records-custody and chain-of-custody rules that name a human custodian. Concrete instances to watch: state court e-filing rules requiring a named human clerk to certify records; HIPAA release-of-information workflows where a designated human must verify identity before disclosure; and state notary statutes (most states still bar remote/automated notarization of certain instruments). If any of these are extended to require a named human attestation on AI-assembled record sets, a thin shield attaches.
Front-desk roles absorbing de facto security and privacy gatekeeping — deciding who gets admitted, what gets disclosed, when to escalate. Employer-side driver is insurer and OSHA workplace-violence pressure (California SB 553 requires a workplace violence prevention plan with designated responsible persons); if such designations land on reception staff by name, the role owns a consequential call.
The limit. Even with every lever, this occupation stays in the low-to-mid 30s, and the levers raise the score of the surviving jobs, not the number of them. The likeliest outcome is a small residual role at a somewhat higher score alongside a very large headcount reduction. There is no plausible route to a trust premium: buyers do not pay extra for a human to key an invoice, and no professional body exists to create licensure.
| New York-Newark-Jersey City, NY-NJ | 136,580 | $47,670 +6% |
| Los Angeles-Long Beach-Anaheim, CA | 112,270 | $48,100 +7% |
| Chicago-Naperville-Elgin, IL-IN | 66,840 | $46,330 +3% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 61,160 | $44,730 -1% |
| Dallas-Fort Worth-Arlington, TX | 52,740 | $44,060 -2% |
| Houston-Pasadena-The Woodlands, TX | 50,040 | $40,370 -10% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 45,660 | $47,060 +5% |
| Atlanta-Sandy Springs-Roswell, GA | 40,710 | $43,330 -4% |
| Boulder, CO | 1,510 | $59,840 +33% |
| San Jose-Sunnyvale-Santa Clara, CA | 13,470 | $58,840 +31% |
| Greeley, CO | 1,570 | $56,030 +24% |
Ford · Central Oregon local governments · UBS · Chubb · Lufthansa · Government of Kazakhstan · PwC · University of Washington
NPR reports that courts in China have ruled in favour of workers dismissed due to AI-driven automation, amid continued job displacement anxiety.
The Times of Central Asia reports Kazakhstan is extending its AI-driven e-government platform into employment and social benefits services.
HR Executive reports Ford rehired workers after AI-related layoffs, prompting HR leaders to reconsider how automation-driven job cut decisions are made.
CNET Japan reports that Ford, after claiming AI would replace half of white-collar roles, quietly rehired 350 skilled technicians.
The Source reports that local government agencies in Central Oregon are using AI chatbots to assist with public sector administrative work.
L'Usine Digitale reports PwC will progressively deploy Anthropic's Claude Code and Cowork tools across its 364,000 employees.
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.