EXPOSED
The real threat here isn't chatbots — it's the CNC and automation curve that has been shrinking this occupation for decades, now accelerated by AI-assisted CAM programming, adaptive toolpath generation, and automated in-machine probing and inspection. What resists is the physical half: fixturing odd parts, dialing in a stubborn setup, hearing a chatter problem and fixing it, changing worn tooling, and running short-run or prototype jobs where writing the program takes longer than cutting the part. Tenders who mostly load, unload, and watch are the most exposed tier; setters who own setup, metrology, and process troubleshooting hold on much longer.
Part 2020 shock, part continued decline in the years since.
Median pay $43,210 → $52,800 -2.2% 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
-14.4% 13,800 → 11,800 on the projections basis
Exposed, and shrinking
Both signals point the same way: the tasks are largely automatable and the BLS projects -14.4% 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.
~1,100 openings a year on average, including replacing people who leave.
MillerScalperBroacherOperatorProfilerSetup ManFile CutterGear RollerLever MillerTooth CutterWheel CutterBarrel RiflerMill OperatorGrind OperatorShift OperatorRouter OperatorShaper OperatorGrinder OperatorMachine OperatorMilling OperatorScalper OperatorProfiler OperatorFeed Mill OperatorKeyseater Operator
Holding it up: embodiment . Weakest point: liability shield .
Mixed — a routine tier and a judgment tier Writing and proving out the program, offsetting for tool wear, and running the second part of a 500-piece order are all things a CAM package plus a probe cycle already does or will do soon — but indicating a casting in a vise to half a thou, shimming a warped plate, and deciding a horn-sounding chatter means feed up not down still needs a person at the machine, which is why this sits at 10 rather than down with pure tenders.
Hands-on in uncontrolled environments You are standing at the machine in coolant mist and chips, hauling vises and angle plates, reaching into the enclosure to change an endmill and re-touch off, deburring and gauging parts with mics and bore gauges — 15 reflects a shop floor that is physically demanding but still a fixed, enclosed, guarded workcell rather than a field or a rooftop.
No licence, no signature requirement There is no state licence to run a mill; a shop can put someone on a Bridgeport or a VMC with in-house training, and when a part goes out of tolerance the liability lands on the shop's quality system and its AS9100/ISO certification, not on your personal ticket — hence a 1, with the only sliver being customer-mandated operator qualification records.
Executes defined procedures on defined inputs Most of the shift is governed by the traveler, the setup sheet, the tooling list, and the inspection frequency, and the calls you do own — bumping a feed override, scrapping a part rather than reworking it, stopping a run when a tool breaks — are real but bounded and reviewable by the next inspection, which is what separates a 6 from the discretion of a process engineer signing off on the method.
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 (10/20 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 (6/20) 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 10 of this occupation's 35 points (29%).
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.
Sheet Metal Workers SAFE
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 48/100, still EXPOSED.
Genuine two-tier structure: if lights-out cells and robotic load/unload absorb the tending tier, the residual occupation is setup, fixture design, metrology interpretation and chatter/thermal troubleshooting on short-run and prototype work — the part of the job where AI-generated CAM still needs a human to prove out the first article. Watchable signal: BLS/employment mix shifting toward 'CNC setter' and 'machinist-programmer' titles while 'machine tender' postings disappear.
Named-operator traceability rather than licensure: AS9100/NADCAP and FDA 21 CFR 820 audits already require identified personnel to sign off first-article inspection and process qualification records. If prime contractors or FAA/FDA auditors explicitly refuse AI-generated in-machine probing data as the sole basis for FAI acceptance and require a named human to certify the dimensional report, this rises from near-zero. Watchable: AS9102 FAI form revisions, NADCAP audit checklists addressing automated inspection.
As the setter's remaining calls become the expensive ones — scrap or rework a $40k forging, accept a borderline GD&T deviation, stop a run on a tool-wear signature — and as those calls stop being buffered by a supervisor in thinner-staffed cells, ownership concentrates. Watchable: MRB (Material Review Board) authority formally delegated to lead setters in quality manuals.
Growth in aerospace/defense and medical hardware work on large, thin-walled or exotic-alloy parts (titanium, Inconel, castings with variable stock) where fixturing and workholding cannot be standardized and each setup is bespoke; robot cells remain economic only above a lot-size threshold. Watchable: DoD/DPA industrial-base funding aimed at forging, casting and large-structure machining capacity.
The limit. No realistic route to a trust premium: buyers of machined parts pay for dimensional conformance to print, not for human hands, and no customer specifies human-cut parts. Liability shield is capped well below professional-licensure levels — machining has no state board, no personal malpractice exposure, and quality signatures sit with the firm's quality organization rather than the operator. Even with every lever, this occupation stays in the exposed-to-mixed band and continues shrinking in headcount; the levers change who survives inside it, not the total.
| Los Angeles-Long Beach-Anaheim, CA | 660 | $54,090 +2% |
| Chicago-Naperville-Elgin, IL-IN | 470 | $63,360 +20% |
| Detroit-Warren-Dearborn, MI | 230 | $56,970 +8% |
| San Diego-Chula Vista-Carlsbad, CA | 230 | $59,050 +12% |
| Elkhart-Goshen, IN | 190 | $59,040 +12% |
| New York-Newark-Jersey City, NY-NJ | 190 | $51,900 -2% |
| San Francisco-Oakland-Fremont, CA | 180 | $58,360 +11% |
| Grand Rapids-Wyoming-Kentwood, MI | 150 | $47,510 -10% |
| Seattle-Tacoma-Bellevue, WA | 110 | $105,430 +100% |
| Louisville/Jefferson County, KY-IN | 90 | $66,440 +26% |
| Chicago-Naperville-Elgin, IL-IN | 470 | $63,360 +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 35. 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.