COOKED
The core work — reading clinical documentation and assigning ICD-10-CM/PCS, CPT and DRG codes, abstracting charts into registries, scrubbing claims, and checking records for completeness — is text-in/label-out pattern work that computer-assisted coding plus LLMs already do at production quality, with humans increasingly auditing rather than coding. What persists is the ambiguous 10%: complex inpatient DRG assignment, clinical documentation improvement queries to physicians, denial and appeal narratives, and release-of-information decisions where HIPAA exposure attaches to a named person. Credentials (RHIT, CCS, CPC) are employer-required and payer-relevant but are not state licensure, so the regulatory shield is thin.
Headcount grew steadily across the period.
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.
BLS projection, 2024–2034
+7.1% 194,800 → 208,600 on the projections basis
Exposed, but growing
AI can already do a lot of these tasks, and the BLS still expects +7.1% more of these jobs by 2034. Demand for the output is growing faster than the work is being automated away — the mechanism BLS gives for software developers, and the combination people most often misread as an error.
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.
~14,200 openings a year on average, including replacing people who leave.
CoderFee CoderMedical CoderMedical BillerMedical ScribeInpatient CoderInsurance CoderDisability RaterOutpatient CoderCoding ConsultantCoding SpecialistMedical Bill CoderMedical Biller CoderMedical Record CoderMedical Billing CoderMedical Records ClerkCertified Medical CoderMedical Insurance CoderHealth Information ClerkHealth Information CoderMedical Claims ProcessorMedical Record AssistantHealth Records TechnicianMedical Coding Specialist
Holding it up: judgment & accountability . Weakest point: embodiment .
Core tasks are already automatable Assigning ICD-10-CM codes from a discharge summary, abstracting tumor-registry fields, and running claim scrubbers are all structured text-to-label mappings that CAC engines already pre-code at 80-90% accuracy on outpatient charts, leaving the coder as a confirm-or-override reviewer; the 5 rather than 0 reflects that multi-comorbidity inpatient DRG sequencing and querying an attending for specificity on 'sepsis vs. bacteremia' still require reading the whole chart against Coding Clinic guidance.
Fully desk- and screen-based The job is a workstation, an EHR, an encoder, and increasingly a home office — the only physical residue is pulling or scanning legacy paper charts, purging shadow files, and walking a release request to a department, which is why this sits at 3 rather than 0.
Certification preferred, not legally required RHIT, CCS and CPC are AHIMA/AAPC certifications that employers and payers demand but no state licenses you, so nothing legally requires a credentialed human to touch a claim; the 6 reflects that upcoding exposes you personally under the False Claims Act and improper disclosure names you in a HIPAA breach report, which is real accountability without the practice-act protection a nurse or RT gets.
Meaningful discretion Choosing a principal diagnosis when two conditions both meet 'after study,' deciding whether a subpoena is valid enough to release psychotherapy notes, and writing an appeal that reframes medical necessity are genuine calls made against official guidelines rather than invented from scratch, which puts this at the low end of real discretion — the coding rules exist, you're interpreting them, and a compliance auditor reviews you.
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 (6/20) 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 (7/20) 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 17 of this occupation's 25 points (68%).
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.
No occupation passed every test: close enough to medical records specialists on skills and subject matter, at least 10 points more resistant, no big jump in training, no new licence, no pay cut, and not shrinking on its own. That happens for 223 of the 654 occupations here that aren't SAFE, and it is worth stating plainly rather than leaving the section off.
The usual reason is that exposure travels with the skill profile. The jobs most similar to yours tend to be exposed for the same reasons yours is, so the near neighbours don't clear the gap — and the ones that do are a different kind of work, not a transfer of what you already know. Read that as a limit of this method, not a verdict that you're stuck: it only compares whole occupations, and it cannot see specialisation, industry, or anything you'd bring that isn't in a federal skill survey.
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 39/100 — EXPOSED.
Task-mix shift is genuinely two-tiered here: if autonomous coding takes the routine outpatient/E&M tier, the residual job is complex inpatient DRG/MS-DRG assignment, CDI physician queries, and denial-appeal narrative construction. Watch for job postings retitled from 'Coder' to 'CDI Specialist' or 'Coding Auditor/Denials Analyst' and for AHIMA/ACDIS certification volume shifting toward CCDS/CDIP.
CMS or OIG guidance under the False Claims Act requiring a named credentialed coder attestation on AI-generated claims — the mechanism to watch is DOJ FCA settlements over autonomous-coding upcoding (already a live theory against EHR/CAC vendors, cf. eClinicalWorks and Practice Fusion settlements), which would push payers and hospital compliance to demand a human sign-off of record on high-DRG and modifier-bearing claims.
If hospital compliance programs formalize the coder as the designated reviewer who can override AI-assigned DRGs and must document rationale for RAC/MAC audit defense — i.e., the role becomes the audit-trail owner rather than the producer — accountability under ambiguity concentrates. Watch for internal coding-compliance policies naming an 'AI output reviewer of record' and payer audit protocols demanding that name.
State licensure (not just certification) for health information management, or a state privacy statute naming an individual release-of-information custodian personally liable for improper PHI disclosure; HHS OCR enforcement plus state laws like Washington's My Health My Data (private right of action) create the pressure, but no state currently licenses HIM staff.
The limit. No plausible route to a meaningful trust premium: the buyer is a payer or hospital revenue cycle, and no patient chooses a human coder. Embodiment is structurally near-zero and fully remote-capable. Even with both liability levers landing, the headcount story is unaffected — a shield that requires one attesting coder per hospital does not preserve 194,720 jobs; it preserves the senior audit tier while the routine tier disappears.
| New York-Newark-Jersey City, NY-NJ | 8,090 | $61,190 +20% |
| Los Angeles-Long Beach-Anaheim, CA | 7,240 | $59,110 +16% |
| Dallas-Fort Worth-Arlington, TX | 5,210 | $52,140 +2% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 4,530 | $49,010 -4% |
| Chicago-Naperville-Elgin, IL-IN | 4,510 | $57,970 +13% |
| Houston-Pasadena-The Woodlands, TX | 3,680 | $51,250 +0% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 3,260 | $49,700 -3% |
| Seattle-Tacoma-Bellevue, WA | 2,830 | $65,530 +28% |
| San Jose-Sunnyvale-Santa Clara, CA | 920 | $86,890 +70% |
| Vallejo, CA | 130 | $82,230 +61% |
| Sacramento-Roseville-Folsom, CA | 1,340 | $75,150 +47% |
Health New Zealand is reported to have implemented AI scribe tools in hospitals to document clinical consultations.
STAT reports that Abridge is adding live prior authorization capability to its AI clinical documentation tool in a partnership with Highmark Health.
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.