EXPOSED
Almost nothing in this job is screen or text work — you hold a die grinder, buffing wheel, or file against a real part and feel when the surface is right, which language models cannot touch. The threat is not AI but capital equipment: CNC machining, robotic sanding cells, and vibratory/media finishing already absorb high-volume, repeatable deburring and polishing work, leaving hand work concentrated in low-volume, irregular, weld-repair, and cosmetic-critical parts. No license, no client relationship, and little decision authority means there is no regulatory or trust moat — only the physical difficulty of automating variable geometry.
This fall is concentrated in 2020 and has not recovered since.
Median pay $30,600 → $42,660 +11.5% 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
-21.2% 11,800 → 9,300 on the projections basis
Exposed, and shrinking
Both signals point the same way: the tasks are largely automatable and the BLS projects -21.2% 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.
~800 openings a year on average, including replacing people who leave.
FilerBufferBurrerSanderStonerBevelerChipperFettlerGrinderFinisherPolisherBit ShaverHand EdgerDeburr HandGlass EdgerHand BufferHand ReamerHand SanderHand ShaperHand StonerPipe BufferGlass SanderHand GrinderHand Scraper
Holding it up: task resistance . Weakest point: liability shield .
Tasks largely resist digitisation Deburring a weld bead on a one-off fabrication, feathering a cosmetic scratch out of a stainless handrail, or blending a casting flash on a part with no fixture — these are contact-force decisions made through the tool, and the 15 reflects that only the setup paperwork and inspection logging around them is digitisable, not the grinding itself.
Hands-on in uncontrolled environments You work standing at a pedestal buffer or with a die grinder in awkward positions, in noise, abrasive dust, and sparks, on parts that arrive at varying orientations — the 15 rather than 19 is because it's usually a fixed bench or shop floor rather than scaffolding, confined spaces, or weather.
No licence, no signature requirement No state license, no certification body governs hand finishing; a shop can put you on the wheel after an OSHA 1910.215 wheel-guarding and ring-test briefing, and rework or scrap comes out of the shop's cost, not your credential.
Executes defined procedures on defined inputs You call grit progression and when to stop against a print tolerance, Ra spec, or a visual boundary sample, and if you're near the low limit you flag it — the discretion is real but bounded by a spec somebody else wrote and QC verifies.
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 (15/20 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 (3/20) is whether buyers specifically pay for a person. Judgment and accountability (5/20) 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 39 points (23%).
Embodiment (15/20) 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.
Furniture Finishers 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 52/100, still EXPOSED.
As robotic sanding cells and vibratory finishing absorb the repeatable high-volume work, the residual hand job concentrates on one-off weld-repair blending, castings with variable geometry, and cosmetic-critical Class A surfaces where fixturing cost exceeds the part value. This is a genuine two-tier occupation and the surviving tier is the harder one.
Named-operator sign-off in aerospace finishing: FAA Part 145 repair station process specs and OEM process specifications (e.g. blend-limit and blade-repair procedures on turbine airfoils) requiring a qualified, individually certified operator stamp on hand-blend operations, with NADCAP audit of that qualification record. This attaches a traceable human name to the work but not personal legal liability in the professional-license sense.
When hand work is the last station before scrap/accept on high-value parts, the polisher owns the call on whether a blend stays within remaining-wall or blend-limit tolerance — a several-thousand-dollar decision made by feel. Formalizing that as an operator accept/reject authority in the work instruction, rather than routing to an inspector, raises this.
Growth in aerospace/turbine MRO and defense shipyard repair work, where the part is in situ, geometry is deviated from nominal, and access is confined — Navy submarine and carrier maintenance backlogs are already pulling this kind of hand finishing in. Robotic cells cannot be brought to the workpiece.
Narrow route only: luxury and restoration niches — hand-polished watch cases, high-end firearms, custom auto and motorcycle bodywork, museum metals conservation — where buyers pay explicitly for hand finish. This applies to a very small slice of the 10,510 and is not a moat for production shops.
The limit. Even with every lever, this occupation stays capital-displaced rather than AI-displaced. The binding constraint is the cost curve of robotic finishing cells and force-controlled end effectors, which no licensing or trust mechanism touches. There is no realistic path to a professional license for hand grinding, so liability_shield cannot rise above a documented-operator-stamp regime. Headcount decline can continue even while per-worker resistance scores rise.
| Los Angeles-Long Beach-Anaheim, CA | 880 | $44,660 +5% |
| Phoenix-Mesa-Chandler, AZ | 490 | $38,280 -10% |
| New York-Newark-Jersey City, NY-NJ | 450 | $51,000 +20% |
| Detroit-Warren-Dearborn, MI | 280 | $46,960 +10% |
| Houston-Pasadena-The Woodlands, TX | 210 | $38,920 -9% |
| Dallas-Fort Worth-Arlington, TX | 200 | $40,250 -6% |
| Riverside-San Bernardino-Ontario, CA | 190 | $42,860 +0% |
| Elkhart-Goshen, IN | 150 | $37,470 -12% |
| Virginia Beach-Chesapeake-Norfolk, VA-NC | 130 | $65,540 +54% |
| St. Louis, MO-IL | 70 | $58,950 +38% |
| Madison, WI | 60 | $56,160 +32% |
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 39. 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.