SAFE
The job is walking the plant: pulling grab samples, backwashing filters, unclogging pumps and screens, changing chlorine cylinders, entering confined spaces, and hand-adjusting valves when SCADA and reality disagree. AI and automation are already eating the logging, trend analysis, and compliance-report drafting — but every state requires a licensed operator of record whose certificate number goes on the discharge monitoring report, and violations carry personal and criminal exposure. Automation has shrunk crews per plant for thirty years and will keep doing so; it has not removed the licensed human who has to be on site when the influent turns.
Roughly flat across the period, with year-to-year wobble.
Median pay $47,760 → $60,020 +0.5% 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
-6.5% 132,400 → 123,800 on the projections basis
Hard to automate, but shrinking anyway
The work resists current AI, yet the BLS projects -6.5% by 2034. Whatever is shrinking this occupation, the evidence does not point to automation — demand, demographics, offshoring and industry decline all shrink jobs that no machine could do.
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.
~10,700 openings a year on average, including replacing people who leave.
FiltererBasin TenderWater PumperFilter TenderPlant OperatorSewer OperatorUtility WorkerWater EngineerWater FiltererWater OperatorRelief OperatorUtility OperatorWatershed TenderDisposal OperatorFiltration OperatorWater Pump OperatorWaterworks EmployeeWaterworks OperatorTreatment SpecialistWaste Water OperatorWater Plant OperatorPurification OperatorSewage Plant OperatorWater System Operator
Holding it up: embodiment . Weakest point: trust premium .
Tasks largely resist digitisation SCADA already trends turbidity, doses coagulant on flow-pace, and auto-generates the monthly operating report — but jar testing a changed raw water, rodding a grit channel, rebuilding a chlorine feed regulator, and calibrating a DO probe against a Winkler titration are all hand work, which is why this lands at 14 and not 18: the paperwork half of the shift is genuinely being automated.
Hands-on in uncontrolled environments A 17 reflects that most of the shift is outdoors and in the wet: climbing clarifier bridges in ice, entering permit-required confined spaces like wet wells and digesters with a four-gas meter, handling one-ton chlorine cylinders and sodium hypochlorite, and pulling lift-station pumps in a manhole — conditions no fixed robot is sited for, though the control room console keeps it off 20.
Licensed human required and personally liable Every state operates a tiered certification scheme under SDWA/CWA primacy (Grade I–IV or A–D), the operator of record's certificate number goes on DMRs and consumer confidence reports, and falsifying or knowingly bypassing carries criminal liability under CWA 309(c) — Flint-era prosecutions of operators are the reason this is 16 rather than a nominal licence at 11.
Exists to be accountable for ambiguous calls When influent hits a hauled-waste slug, a nitrification crash drops effluent ammonia out of permit, or a storm forces the call between blending and a sanitary sewer overflow, the operator on shift decides in minutes with no procedure covering it — 14 rather than higher because most days run inside established setpoints and the chief operator or engineer owns process changes.
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 (14/20 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 (16/20) is whether the law requires a licensed human to sign. Trust premium (7/20) is whether buyers specifically pay for a person. Judgment and accountability (14/20) 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 37 of this occupation's 68 points (54%).
Embodiment (17/20) 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 76/100, still SAFE.
Task-mix shift is real here: logging, trend analysis, and DMR drafting are the routine tier and are already being automated, leaving the storm-event, influent-upset, PFAS/lead-rule-compliance, and cyber-incident judgment tier. If EPA's PFAS NPDWR and revised Lead and Copper Rule Improvements force more frequent non-routine treatment decisions and public-notification calls, the surviving job is mostly consequential calls under ambiguity.
Post-crisis criminal enforcement precedent extending Flint-style prosecutions (Michigan charged operators and supervisors) or the Clean Water Act 33 USC 1319(c) knowing-endangerment provisions to cases where an operator accepted an automated setpoint — making personal certificate exposure the explicit reason a human stays in the loop, plus cyber rules (AWIA 2018 risk-and-resilience certifications) naming a certified individual.
EPA or state primacy agencies tightening operator-in-responsible-charge rules so that remote/centralized monitoring cannot substitute for a certified operator physically present per shift — i.e., states rewriting minimum staffing rules under 40 CFR 141 Subpart Y/ABC standardized certification guidelines to explicitly bar AI or remote-only coverage. Several states (e.g., Texas TCEQ, California SWRCB T2-T5 grades) already set hours-of-attendance minimums; codifying that AI-generated adjustments require a graded operator's signature would harden it.
Aging infrastructure with deferred capital (ASCE grades US wastewater D+) means more emergency manual intervention — confined-space entry, bypass pumping, chlorine cylinder swaps — in plants where instrumentation is unreliable. Also OSHA 1910.146 permit-required confined space entry requires human attendants and entrants; any tightening of gas-detection or attendant rules raises the floor.
The limit. trust_premium has no plausible route: ratepayers do not choose their utility, cannot perceive who operated the plant, and will never pay a premium for a human operator. Municipal budget pressure pushes the opposite way — toward contract operations (Veolia, Jacobs) and remote-monitoring consolidation of small systems. The realistic ceiling is roughly 72-75, and headcount can still fall sharply even as the per-plant licensed role stays legally mandatory: the license protects the position, not the number of positions.
| New York-Newark-Jersey City, NY-NJ | 3,690 | $83,320 +39% |
| Phoenix-Mesa-Chandler, AZ | 2,570 | $63,660 +6% |
| Los Angeles-Long Beach-Anaheim, CA | 2,230 | $82,460 +37% |
| Houston-Pasadena-The Woodlands, TX | 1,980 | $51,940 -13% |
| Dallas-Fort Worth-Arlington, TX | 1,910 | $51,780 -14% |
| Chicago-Naperville-Elgin, IL-IN | 1,770 | $79,290 +32% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 1,650 | $66,680 +11% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 1,630 | $73,600 +23% |
| San Francisco-Oakland-Fremont, CA | 1,040 | $125,160 +109% |
| San Jose-Sunnyvale-Santa Clara, CA | 460 | $108,190 +80% |
| Vallejo, CA | 200 | $106,150 +77% |
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 68. 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.