SAFE
Much of the shift is instrument monitoring, log-keeping, and procedure-following that automation already handles or could — reactor control systems are heavily instrumented and self-regulating by design. What protects the job is NRC 10 CFR Part 55: a licensed human operator must be at the controls, personally accountable for reactivity manipulations, and answerable in a way no algorithm can be. Add plant walkdowns, valve lineups, and abnormal-transient judgment where procedures conflict, and displacement risk is low even as headcount per unit slowly falls.
Roughly flat across the period, with year-to-year wobble.
Median pay $100,530 → $122,890 -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
-15.3% 5,700 → 4,900 on the projections basis
Hard to automate, but shrinking anyway
The work resists current AI, yet the BLS projects -15.3% 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.
~400 openings a year on average, including replacing people who leave.
Control OperatorNuclear OperatorControl Room AgentControl Room OfficerControl Room OperatorNuclear Unit OperatorReactor Operator (RO)Unit Reactor OperatorPower Reactor OperatorNuclear Control OperatorNuclear Reactor OperatorLicensed Reactor OperatorNuclear Plant Operator (NPO)Nuclear Control Room OperatorNuclear Power Reactor OperatorNuclear Station Operator (NSO)Nuclear Supervising Operator (NSO)Licensed Nuclear Control Room OperatorNuclear Control Room Non-Licensed OperatorNuclear Power Plant Process Control Worker
Holding it up: liability shield . Weakest point: trust premium .
Mixed — a routine tier and a judgment tier Steady-state load-following, rod position adjustment, and turbine synchronization are already digitally controlled in newer plants, but transient response — diagnosing a conflicting alarm cascade, deciding whether to trip or ride out a feedwater upset, executing EOP transitions when symptoms don't match the flowchart — keeps roughly half the job resistant, which is why this lands at 12 rather than in the safe teens.
Some physical or field component Most of the shift is in a seated control-room panel or simulator, but operators still perform surveillance walkdowns in the containment and turbine building, manually stroke and tag valves for lineups, hang clearances, and physically operate local controls during loss-of-control-room scenarios — real hands-on work in radiological and high-noise areas, just not the majority of hours, hence 11.
Licensed human required and personally liable An NRC Reactor Operator or Senior Reactor Operator licence under 10 CFR Part 55 is issued to a named individual after a written exam, operating test, and medical certification, requires biennial renewal and continuing requalification, and can be revoked or criminally enforced against that person for a deliberate misconduct violation — you cannot legally manipulate the controls without it, which is close to the ceiling at 19.
Exists to be accountable for ambiguous calls During an unanalyzed event the licensed operator invokes 10 CFR 50.54(x) authority to depart from licence conditions to protect public health, and routinely makes trip/no-trip and safety-system-actuation calls under minutes of time pressure with core damage as the downside — 17 reflects that the consequences are civilization-scale even though EOPs and AOPs constrain much of the decision space.
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 (19/20) is whether the law requires a licensed human to sign. Trust premium (11/20) is whether buyers specifically pay for a person. Judgment and accountability (17/20) 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 47 of this occupation's 70 points (67%).
Embodiment (11/20) 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 78/100, still SAFE.
Genuine two-tier structure: if log-keeping, surveillance-test scheduling, procedure retrieval, and trend monitoring are absorbed by digital I&C and computerized procedure systems, the residual shift is abnormal-transient diagnosis, conflicting-procedure resolution, and reactivity manipulation under degraded instrumentation. Headcount per unit falls but the remaining job is harder to automate. Watch for utilities deploying computer-based procedures under NUREG-0711 human-factors review.
If NRC or INPO guidance formally assigns the licensed operator the duty to accept, reject, or override recommendations from digital/AI-based control and diagnostic systems — a named human as the deciding authority for automation disagreement during transients, analogous to the operator-authority provisions in digital I&C upgrade licensing (e.g. Diablo Canyon/Hatch DCS reviews) — the role's ambiguity ownership becomes explicit rather than implicit.
If the fleet shift toward life-extension of aging units (80-year license renewals at Peach Bottom, Surry, Oconee) increases the frequency of degraded-equipment walkdowns, manual valve lineups, and local operation of components with failed remote actuators, the unpredictable-physical share of the shift rises.
NRC codifying operator-licensing requirements for advanced/SMR designs rather than granting reduced-staffing exemptions — i.e. if the Part 50.54(m) minimum-staffing rulemaking and the 10 CFR 53 Part 53 licensing framework for non-light-water reactors end up requiring a licensed senior reactor operator at the controls per unit regardless of autonomous control credit, and explicitly bar AI systems from crediting as an operator. The countervailing case is already visible: NuScale's approved 6-units-per-control-room staffing and the ADVANCE Act's push toward reduced staffing.
The limit. Trust premium has no realistic route — ratepayers and grid operators do not choose a plant because humans run it, and no buyer-facing preference exists to monetize. The dominant risk here is not AI capability but NRC-sanctioned staffing reduction: the ADVANCE Act and SMR multi-unit control room approvals cut operators per megawatt without touching any of these dimension scores, so the occupation can stay 'safe' per-worker while total employment falls.
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 70. 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.