SAFE
The job is walking track, opening signal cases and switch machines in weather, testing relays and track circuits, adjusting point detection, and splicing cable — none of which a language model can touch and none of which current robotics can do on live right-of-way. AI will absorb the paperwork side: test records, defect logs, scheduling of periodic inspections, and remote diagnostics from wayside monitoring that flags a failing circuit before you arrive. What persists is the FRA-qualified human who physically performs and certifies signal tests and decides whether a track can be released to traffic.
Headcount grew steadily across the period.
Median pay $73,890 → $92,460 +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
+1.7% 8,700 → 8,800 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +1.7% more of these jobs by 2034. Note that safe does not mean well paid — several of the fastest-growing resistant occupations are among the lowest paid on the register.
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.
~800 openings a year on average, including replacing people who leave.
SignalmanSignal ExpertSignal WorkerGate TechnicianSignal MechanicSignal InspectorRegister RepairerSignal MaintainerSignal TechnicianRail Signal DesignerRail Signal MechanicRail Signal InspectorRail Track MaintainerSignal Systems ExpertTest Signal TechnicianTraffic Signal InspectorTrain Control TechnicianCommunications MaintainerRailway Signal TechnicianRailroad Signal MaintainerLight Rail Signal TechnicianSignal Maintenance TechnicianElectric Track Switch MaintainerSignal System Testing Maintainer
Holding it up: embodiment . Weakest point: trust premium .
Tasks largely resist digitisation Diagnostics can tell you a track circuit is drifting, but nobody has built a machine that pulls a switch machine cover in sleet, files contacts, adjusts point detection to a quarter-inch obstruction test, and re-cranks it under a fouling time window — the ballast, the varied hardware vintages, and the cable splicing keep this at 16 rather than mid-band.
Hands-on in uncontrolled environments You work on live right-of-way with train movements around you, in bungalows and cases at grade crossings, in mud and heat and ice, climbing to signal heads and digging to cable — an environment no controlled-cell robot handles, which is why this sits at 19 instead of 13.
Licensed human required and personally liable There's no state licence, but 49 CFR 236 requires you to be FRA-qualified for the specific tests you perform and your name goes on the test record the FRA inspector reads — a personal, auditable federal obligation that lands at 12 rather than 18, since the railroad's signal department, not you, holds the system-level responsibility.
Meaningful discretion When you decide a switch's point detection is within tolerance, whether to place a track out of service, or how to protect a crossing running in flash mode, you're making calls with train movements and crossing motorists on the other side — real discretion under 236 rules, though the rules and test procedures themselves are written for you, holding it at 13.
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 (16/20 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 (12/20) 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 (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 32 of this occupation's 67 points (48%).
Embodiment (19/20) 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 79/100, still SAFE.
A minimum-signal-crew or minimum-qualification provision in a national BRS agreement or a state rail safety statute (analogous to the two-person crew laws enacted in CA, WA, AZ, NV and the FRA 2024 crew size rule) requiring an on-ground qualified signalman for switch/interlocking work rather than a contractor with remote support.
Task-mix shift is genuine here: wayside condition monitoring and automated defect logging absorb the routine periodic-inspection tier, leaving intermittent-fault diagnosis, cutover work on new interlockings, and storm/derailment restoration — the tier where the failure is not what the sensor said it was. Score rises only as the routine tier leaves, and only if headcount is not cut proportionally.
FRA tightening 49 CFR Part 236 Subpart A qualification/certification rules so that any signal test — including those prompted or pre-diagnosed by wayside/PTC monitoring — must be performed and personally certified by a named FRA-qualified signal employee, with remote or algorithmic verification explicitly disallowed as a substitute. FRA has an open track record of rulemaking here (Part 236 revisions, PTC-era test requirements) and BRS (Brotherhood of Railroad Signalmen) has actively pushed against remote/automated test substitution in agreement negotiations and FRA comment dockets.
Release-to-traffic authority being formalized as a personally logged decision — an FRA or carrier rule naming the individual signal employee who authorizes restoration of a signal or switch to service after failure, with that name attached in NTSB/FRA accident investigations. This is the pattern already visible in post-incident FRA findings after signal-related derailments.
The limit. Embodiment at 19 and task_resistance at 16 are effectively at ceiling; the composite has little room. Trust premium has no plausible route — the buyer is a Class I railroad or transit agency purchasing compliance, not a human touch, and no end passenger ever chooses a signal repairer. The real risk to this occupation is headcount compression from remote diagnostics and contracting-out, not task automation, and no dimension here captures that.
| New York-Newark-Jersey City, NY-NJ | 1,080 | $94,530 +2% |
| Seattle-Tacoma-Bellevue, WA | 170 | $121,710 +32% |
| Seattle-Tacoma-Bellevue, WA | 170 | $121,710 +32% |
| New York-Newark-Jersey City, NY-NJ | 1,080 | $94,530 +2% |
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 67. 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.