COOKED
The core of this job — keying text from drafts or dictation, formatting reports and correspondence, applying templates and styles, proofreading for spelling and grammar — is precisely what speech-to-text and generative models already do at production quality and near-zero marginal cost. There is no licensure, no physical component beyond sitting at a keyboard, and no client who is paying for this specific person's presence. Employment has already fallen by roughly 90% from its 1990s peak, and the remaining 35,000 jobs are largely in legal, medical, and government offices where formatting rules and records handling are still tolerated as human work.
Part 2020 shock, part continued decline in the years since.
Median pay $40,340 → $49,280 -2.3% 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
-36.1% 40,000 → 25,600 on the projections basis
Exposed, and shrinking
Both signals point the same way: the tasks are largely automatable and the BLS projects -36.1% 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.
~2,200 openings a year on average, including replacing people who leave.
TypistAddresserNotereaderScript GirlTranscriberClerk TypistLegal TypistOffice ClerkStenographerPolicy TypistScript WorkerMedical TypistStencil TypistWord ProcessorBordereau ClerkClerk SpecialistContinuity ClerkData TranscriberOffice AssistantOffice AssociateTranscriptionistDictaphone TypistEdiphone OperatorOffice Technician
Holding it up: task resistance . Weakest point: liability shield .
Core tasks are already automatable Transcribing dictation, keying text into templates, applying styles and running spell-check are tasks where Whisper-class ASR and off-the-shelf document automation already hit or beat human accuracy per page, which is why this sits at 2 rather than 7 — even the residual work of merging revisions and building tables of contents is native software functionality, not a human judgment gap.
Fully desk- and screen-based Sitting at a workstation with a keyboard, headset, and occasionally a scanner or copier is the whole physical footprint; the 2 rather than 0 acknowledges retrieving paper files and stuffing envelopes in legal and clinical offices, not any task requiring hands in an uncontrolled space.
No licence, no signature requirement There is no state licence, no certification board, and no signature line — a typo in a deposition transcript or discharge summary comes back on the attorney or clinician who signed it, and employers can and do route the same work to an offshore vendor or an ASR pipeline with no credentialing barrier.
Executes defined procedures on defined inputs Deciding whether an inaudible passage gets bracketed, which template applies to a motion, or whether to query the dictator on an ambiguous drug name is real but bounded discretion inside a style manual, and every substantive call escalates to the originating attorney or physician — that keeps it at 2 rather than 7.
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 (2/20 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 (0/20) is whether the law requires a licensed human to sign. Trust premium (2/20) is whether buyers specifically pay for a person. Judgment and accountability (2/20) 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 4 of this occupation's 8 points (50%).
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 24/100, still COOKED.
Task-mix shift within the surviving legal/medical/court niche: if ASR handles the raw keying, the remaining work concentrates in certified transcript preparation to court-specific format (e.g., federal court transcript format rules, AAERT CET-style verbatim standards), redaction of PII/PHI under HIPAA and FRCP 5.2, and speaker attribution in multi-party audio — tasks where error is consequential and current ASR degrades on crosstalk and accents. This is a genuine two-tier occupation; the judgment tier is small but real.
Buyer-side rules against sending sensitive audio to cloud AI — classified, grand jury, sealed family-court, or attorney-client material handled under protective order — sustaining a small paid-for-human-only segment; visible today in government contracts specifying US-person, cleared transcriptionists.
Court rules or state statutes requiring that a transcript entered into the record be certified by a human transcriptionist/reporter attesting to accuracy — already the norm for court reporters (many states bar AI-only certified transcripts), and could be extended by judicial council rule to transcript-prep staff handling recorded proceedings (as in states using electronic recording plus certified transcribers).
Health-system or state rules requiring a credentialed medical transcriptionist/editor (AHDI CHDS/RHDS) to review and sign off on AI-drafted clinical documentation before it enters the legal medical record, driven by malpractice-insurer requirements after documented ambient-scribe hallucination incidents (Whisper hallucination findings in clinical audio, 2024).
Role redefinition to AI-output verifier of record: if the transcriptionist is named as the person attesting to accuracy of an AI-generated transcript or clinical note, they own the call on ambiguous audio and on what gets redacted, with discoverable consequences.
The limit. Even with every lever, this tops out in the low 20s and only for the fraction of the 35,010 sitting in courts, hospitals, and cleared government work. The generic office typist has no route; the credentialed niche is a different job with a different title, and headcount there is already small and shrinking.
| New York-Newark-Jersey City, NY-NJ | 8,370 | $52,290 +6% |
| Los Angeles-Long Beach-Anaheim, CA | 6,830 | $57,670 +17% |
| San Juan-Bayamon-Caguas, PR | 1,330 | $28,940 -41% |
| Buffalo-Cheektowaga, NY | 1,300 | $47,040 -5% |
| Riverside-San Bernardino-Ontario, CA | 880 | $49,280 +0% |
| San Francisco-Oakland-Fremont, CA | 840 | $79,430 +61% |
| Albany-Schenectady-Troy, NY | 830 | $46,550 -6% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 690 | $47,250 -4% |
| San Francisco-Oakland-Fremont, CA | 840 | $79,430 +61% |
| San Jose-Sunnyvale-Santa Clara, CA | 100 | $64,280 +30% |
| Worcester, MA | 90 | $59,110 +20% |
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 8. 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.