EXPOSED
The core loop — read the spec, set furnace temperature and cycle time, load and unload baskets, quench, test hardness, log the chart — is half physical handling and half parameter work that PLCs, recipe management software and closed-loop pyrometry already do better than a human eye. Loading heavy fixtures, hanging parts, breaking up jams, straightening warped stock and troubleshooting a furnace that isn't holding uniformity still needs a body on a hot shop floor, which is the main thing holding this job in place. The bigger threat isn't a language model, it's capital investment in automated batch lines and the long-run decline in domestic heat treat headcount.
This fall is concentrated in 2020 and has not recovered since.
Median pay $38,250 → $48,750 +2.0% 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
-12.8% 14,800 → 12,900 on the projections basis
Exposed, and shrinking
Both signals point the same way: the tasks are largely automatable and the BLS projects -12.8% 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.
~1,200 openings a year on average, including replacing people who leave.
BaggerBurnerPusherChargerPeelmanAnnealerHardenerReheaterTempererCarbonizerNormalizerTube HeaterBox AnnealerDie HardenerFagot HeaterForge HeaterHeat TreaterRivet HeaterSheet HeaterSteel HeaterStove TenderBillet HeaterCase HardenerFace Hardener
Holding it up: embodiment . Weakest point: liability shield .
Mixed — a routine tier and a judgment tier Recipe entry, ramp-soak-quench cycles and hardness data logging are already handled by PLCs and AMS 2750 pyrometry systems that hold uniformity tighter than any operator's judgment, but the physical half — racking gears, hanging shafts, moving baskets from furnace to quench tank before the part cools out of range, clearing a jam in a mesh belt — is what puts this at 10 instead of the 4 a pure control-panel job would score.
Hands-on in uncontrolled environments You work within feet of a 1700°F furnace door in leathers and a face shield, manhandle fixtures and hot baskets with tongs or a hoist, deal with quench oil fumes and salt bath splash, and chase temperature drift by physically checking thermocouples and burner tuning — uncontrolled enough to earn 14, but it's still a fixed shop floor with fixed equipment rather than a 17-18 field environment.
No licence, no signature requirement No state licence gates this work; the certification that matters — Nadcap or AS9100 heat treat approval, AMS 2750 pyrometry compliance — attaches to the plant and its quality department, not to your name, and when a batch fails hardness testing the customer pursues the shop, so 2 rather than 0 only reflects the signed process charts you initial.
Meaningful discretion The spec sheet dictates temperature, soak time and quench medium, and deviations need engineering sign-off, so most shifts are procedure-following; the 7 covers the calls that are genuinely yours — deciding a furnace isn't holding uniformity and stopping production, judging whether warped stock is salvageable by straightening, adjusting for load density or a part that came in with unexpected prior condition.
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 (2/20) 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 (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 12 of this occupation's 36 points (33%).
Embodiment (14/20) 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 50/100, still EXPOSED.
Genuine two-tier structure: the routine tier (set temp, start cycle, log chart) is the part PLCs and recipe management already absorb; the residual tier is furnace troubleshooting — chasing uniformity failures, burner and atmosphere problems, fixture and jam recovery, diagnosing why a specific alloy came out soft. If headcount consolidates so remaining operators are effectively furnace troubleshooters covering multiple automated lines, the measured day's work becomes harder to automate even with no capability regression.
AMS2750 (aerospace pyrometry) and CQI-9 (AIAG automotive heat treat assessment) already require named, qualified personnel to sign off on TUS/SAT surveys, instrument calibration and load charts. If Nadcap audit criteria or a customer flowdown were tightened to require a certified heat treat operator — not just a metallurgist or quality engineer — to personally attest each production load's chart and any recipe deviation, the shop-floor role acquires a signature that cannot be delegated to the PLC. Watch AMS2750 revisions and Nadcap HT task-group checklists for operator-level certification language.
Task-mix shift plus deviation authority: on lines where recipe execution is fully closed-loop, the human's remaining calls are the ambiguous ones — accept/scrap a load that drifted out of uniformity, decide whether warpage is straightenable or the batch is condemned, stop the furnace on a suspect thermocouple. If a quality system (CQI-9 or an internal MRB procedure) names the operator as the person who owns the disposition and the nonconformance record rather than routing it to engineering, this rises.
Unpredictable-environment content rises if work shifts toward larger, non-repeating loads — big forgings, mixed job-shop lots, induction and flame hardening of oddly shaped parts — where fixturing is bespoke and no automated basket handler pays back. Reshoring of low-volume defense and aerospace forging work would push this direction; high-volume captive automotive lines push the other way.
The limit. No realistic trust-premium route: buyers purchase a hardness spec and a certified chart, and would happily accept a lights-out furnace that meets it. Also note the dominant threat here is capital replacement of the whole line rather than task-level AI, so gains in liability_shield and judgment_accountability protect the role only in shops that survive the consolidation at all — 14,000 workers and shrinking.
| Houston-Pasadena-The Woodlands, TX | 700 | $47,550 -2% |
| Chicago-Naperville-Elgin, IL-IN | 690 | $45,100 -7% |
| Detroit-Warren-Dearborn, MI | 650 | $50,790 +4% |
| Los Angeles-Long Beach-Anaheim, CA | 420 | $49,680 +2% |
| New York-Newark-Jersey City, NY-NJ | 420 | $50,010 +3% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 250 | $49,470 +1% |
| Cleveland, OH | 240 | $50,690 +4% |
| Nashville-Davidson--Murfreesboro--Franklin, TN | 240 | $56,280 +15% |
| Canton-Massillon, OH | 170 | $67,070 +38% |
| Oklahoma City, OK | 40 | $65,640 +35% |
| Albany, OR | 40 | $64,210 +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 36. 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.