EXPOSED
Most of the control-room half of this job — trending process variables, logging readings, adjusting setpoints within envelope, generating shift reports — is already handled by DCS/APC systems and is exactly where AI advanced-control layers are being sold. What keeps headcount is the field half: manual valve lineups, pump swaps, sampling, hunting a leak or a plugged line, and being the human who executes and judges an abnormal startup, shutdown, or emergency response under process-safety rules. Staffing has already been squeezed to a thin crew per unit, so further automation cuts consoles more than it cuts boots on the unit.
Part 2020 shock, part continued decline in the years since.
Median pay $62,550 → $78,120 -0.1% 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
-6.1% 18,100 → 17,000 on the projections basis
Shrinking, but not obviously because of AI
The BLS projects -6.1% by 2034, but at 54/100 this work is only moderately exposed — not the profile of a job current AI can simply do. Occupations shrink for many reasons, and the score does not point at automation as this one's cause.
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.
~1,600 openings a year on average, including replacing people who leave.
DenitratorWash OperatorTower OperatorPlant AssociateProcess OperatorUtility OperatorChemical OperatorLoader TechnicianNitrogen OperatorAmmonia TechnicianProcess TechnicianUtilities OperatorFront Wash OperatorPressure ControllerProcessing OperatorProduction OperatorBucket Wash OperatorProduction TechnicianAmmonia Still OperatorFurnace Plant OperatorHydrogenation OperatorChemical Plant OperatorPharmaceutical OperatorFurnace Utility Operator
Holding it up: embodiment . Weakest point: trust premium .
Mixed — a routine tier and a judgment tier Advanced process control already holds the loops, ramps rates and writes the shift log, but nobody has automated a cold startup from a turnaround, a manual bypass lineup around a fouled exchanger, or diagnosing why a column is flooding when the level transmitter reads fine — that split of the day is what sits it at 12 rather than 6 or 16.
Hands-on in uncontrolled environments Climbing structures to gag a relief valve, breaking flanges for line-breaking permits, pulling grab samples off a hot header in a fire-retardant suit and respirator, and swapping pumps in weather and H2S monitoring range is uncontrolled physical work at 16; it stops short of 20 because a real share of the shift is genuinely spent seated at the console.
Certification preferred, not legally required There is no state operator licence for chemical units the way there is for boilers or wastewater — what you carry is HAZWOPER, RCRA, DOT and unit-specific qualification cards plus your name on the permit and the log, so PSM (29 CFR 1910.119) liability lands on the employer's process-safety program, not on you personally, which is why this is a 7 and not an 11.
Meaningful discretion When a compressor trips at 3 a.m. and you have minutes to decide whether to route to flare, initiate emergency depressurization or ride it out, procedures narrow the choice but do not make it for you, and misreading a runaway exotherm is a fatality event — 13 rather than 16 because the written operating and emergency procedures and a callable engineer or supervisor cover most abnormal cases.
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 (12/20 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 (7/20) is whether the law requires a licensed human to sign. Trust premium (6/20) is whether buyers specifically pay for a person. Judgment and accountability (13/20) 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 26 of this occupation's 54 points (48%).
Embodiment (16/20) 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.
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 71/100 — SAFE.
Genuine two-tier structure: if APC/AI absorbs the trending, logging, and in-envelope setpoint tier, the residual job is abnormal-event diagnosis, field troubleshooting, and turnaround/startup execution — the tier that resists. This raises task_resistance without any new law, but it comes with fewer consoles, so score rises while headcount falls.
OSHA PSM (29 CFR 1910.119) and EPA RMP are being reopened — EPA's 2024 Safer Communities rule already tightened root-cause and safeguard requirements. A revision that names a qualified human operator as the required authority for any deviation outside the safe operating envelope, and requires a named operator signature on lockout/tagout, line-break, and startup/shutdown permits that an AI control layer cannot hold, would make a credentialed human legally load-bearing rather than merely present. CSB recommendations after an AI-supervised-control incident are the most likely trigger.
Formalisation of the operator as the accountable decision-maker for abnormal-situation management — e.g. an ISA/IEC 61511 revision or ASM Consortium practice adopted in site procedures that assigns the shutdown decision explicitly to the console operator and requires documented operator rationale when AI advice is overridden or accepted. This converts diffuse judgment into an owned, auditable call.
Insurer-driven rather than regulator-driven: property/business-interruption underwriters (AIG, Chubb, FM) imposing a written condition that autonomous setpoint changes above a defined severity require operator concurrence logged in the DCS, as a coverage condition. Refinery and chemical underwriting already prices staffing and PSM audit quality.
Already near ceiling at 16. Marginal further rise only if plants push more manual field work onto operators as console roles thin — sampling, valve lineups, and leak-hunting in congested, hot, classified (Class I Div 1) areas where mobile robotics remain uneconomic at 16,610-worker scale.
The limit. Trust premium has no plausible route: the buyer is the plant owner, who is cost-minimising on labour, and no downstream customer of ethylene or chlorine pays extra for a human operator. Realistic ceiling around 66-68, and the task_resistance and embodiment gains are compatible with substantial headcount loss — a smaller, harder-to-automate job is not a safer occupation for the current workforce.
| Baton Rouge, LA | 1,720 | $104,410 +34% |
| Houston-Pasadena-The Woodlands, TX | 1,420 | $106,020 +36% |
| Lake Charles, LA | 740 | $102,310 +31% |
| New Orleans-Metairie, LA | 460 | $92,260 +18% |
| Boston-Cambridge-Newton, MA-NH | 410 | $82,170 +5% |
| Richmond, VA | 340 | $77,930 +0% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 310 | $77,260 -1% |
| San Francisco-Oakland-Fremont, CA | 300 | $111,900 +43% |
| San Francisco-Oakland-Fremont, CA | 300 | $111,900 +43% |
| Houston-Pasadena-The Woodlands, TX | 1,420 | $106,020 +36% |
| Baton Rouge, LA | 1,720 | $104,410 +34% |
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 54. 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.