EXPOSED
The bulk of hours at the national median go to reconciliations, journal entries, transaction tie-outs, variance schedules, tax return prep and workpaper documentation — all structured text-and-number work that AI plus modern ledger software already does at usable quality. What survives is the part where a licensed human signs: the audit opinion, the tax position taken under ambiguous law, the materiality and going-concern call, and the conversation with a client or audit committee about what the numbers actually mean. This title bundles the staff accountant (highly exposed) with the CPA partner and controller (well protected); the modal worker sits closer to the exposed end.
Headcount grew steadily across the period.
Median pay $71,550 → $83,680 -6.4% 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
+4.6% 1,579,800 → 1,652,600 on the projections basis
Exposed, but growing
AI can already do a lot of these tasks, and the BLS still expects +4.6% 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.
~124,200 openings a year on average, including replacing people who leave.
BursarAuditorAccountantTax AuditorCity AuditorAsset AnalystAudit PartnerField AuditorStaff AuditorTax AssociateCounty AuditorTax AccountantTax SpecialistAccount AuditorAudit AssociateBank AccountantCost AccountantFund AccountantMedical AuditorPayroll AnalystPayroll AuditorInternal AuditorRailroad AuditorStaff Accountant
Holding it up: judgment & accountability . Weakest point: embodiment .
Mixed — a routine tier and a judgment tier A 7 reflects that bank and intercompany reconciliations, three-way match testing, sampling selection, depreciation schedules, and Form 1120/1065 preparation from a trial balance are now largely tool-driven, while a residue of hours — walkthroughs of a client's controls, chasing an unexplained accrual through emails and contracts, drafting footnote disclosures for a new revenue arrangement — still requires someone assembling facts nobody has coded yet, which is why this sits at the top of the automatable band rather than at 3.
Fully desk- and screen-based A 2 covers the thin physical tail: inventory observation counts on a warehouse floor at year-end, fixed-asset existence checks, and cash counts at a client site, which for the median accountant is a few days a year against a calendar of ERP screens, Excel, and workpaper software.
Licensed human required and personally liable An 11 sits just inside the licensed band because the CPA licence is legally load-bearing only at the top of the workpaper — the engagement partner signs the opinion and faces PCAOB sanction, state board discipline, and Section 10A exposure, while the staff and senior accountants who do most of the tick-and-tie, and the entire uncredentialled corporate accounting population under a controller, hold no licence that anyone must have on file.
Meaningful discretion A 12 marks the genuine discretion that exists — setting materiality thresholds, deciding whether a contingency is probable and estimable under ASC 450, taking a tax position on economic substance, judging whether management's going-concern plans are realistic — against the fact that most of these calls run through GAAP, IRS regulation, and firm methodology templates with a reviewer above the preparer, so the ambiguous call is escalated rather than owned by the person who found it.
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 (7/20 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 (11/20) is whether the law requires a licensed human to sign. Trust premium (10/20) is whether buyers specifically pay for a person. Judgment and accountability (12/20) 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 33 of this occupation's 42 points (79%).
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.
Personal Financial Advisors 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 57/100, still EXPOSED.
PCAOB/SEC or state boards adopting an explicit rule that AI-generated audit evidence and analytics require a named engagement partner's documented review and sign-off (PCAOB already has an active standard-setting project on the use of technology-assisted analysis, and AS 1201 supervision requirements); plus state board rules that tax positions produced by software must be signed by a licensed preparer with Circular 230 liability attaching personally
Genuine two-tier structure: if reconciliations, tie-outs and return prep are fully automated, the residual measured job becomes materiality judgment, going-concern assessment, fraud-risk skepticism, and audit-committee communication — the tier AI cannot own. This raises the score only for those who remain employed; headcount falls at the staff tier
Expanded ICFR and fraud-detection expectations (post-Wirecard/FTX enforcement, SEC clawback and internal-control rules) pushing more explicitly ambiguous calls onto the named signer, plus any rule making the auditor accountable for evaluating the client's AI-generated accounting estimates
Audit-quality insurers and peer-review programs requiring human attestation that AI outputs were tested, making uninsured AI-only workpapers effectively unusable in litigation defense
Audit committees and lenders specifying a human-partner-reviewed opinion in engagement letters or debt covenants; some private-equity and bank lenders already demand named-partner accountability rather than firm-level attestation
The limit. Every lever concentrates value in the licensed signer. The modal worker in this SOC is a staff accountant whose hours are exactly the automated tier; liability and judgment gains accrue to a shrinking partner/controller layer, so the occupation's overall score can rise while employment in it falls.
| New York-Newark-Jersey City, NY-NJ | 111,930 | $105,650 +26% |
| Los Angeles-Long Beach-Anaheim, CA | 70,640 | $94,360 +13% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 41,470 | $102,140 +22% |
| Chicago-Naperville-Elgin, IL-IN | 41,360 | $82,360 -2% |
| Dallas-Fort Worth-Arlington, TX | 37,450 | $82,620 -1% |
| Boston-Cambridge-Newton, MA-NH | 37,150 | $100,150 +20% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 30,950 | $80,920 -3% |
| Atlanta-Sandy Springs-Roswell, GA | 30,150 | $85,820 +3% |
| San Jose-Sunnyvale-Santa Clara, CA | 13,180 | $121,700 +45% |
| San Francisco-Oakland-Fremont, CA | 25,170 | $110,730 +32% |
| New York-Newark-Jersey City, NY-NJ | 111,930 | $105,650 +26% |
Internal Revenue Service · Goldman Sachs · KPMG · PwC
L'Usine Digitale reports PwC will progressively deploy Anthropic's Claude Code and Cowork tools across its 364,000 employees.
The U.S. Government Accountability Office reported on the IRS's use of artificial intelligence in its operations, including audit selection and taxpayer service functions.
CNBC reports Goldman Sachs is deploying Anthropic's Claude to automate work in accounting and compliance functions.
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.