SAFE
The work is physically embedded in airframes: pulling and terminating wire bundles in bulkheads and wing roots, bench-testing and swapping LRUs, running ramp-side troubleshooting on intermittent faults that no schematic predicts, and torquing connectors in spaces built for hands, not robots. AI will absorb the paperwork layer — fault-code interpretation, manual lookup, wiring-diagram search, logbook drafting — but a certificated human still signs the return-to-service entry and personally carries FAA liability for it. The diagnostic call on an unrepeatable squawk, where the choice is ground the aircraft or release it, is exactly the accountable judgment that doesn't delegate.
This fall is concentrated in 2020 and has not recovered since.
Median pay $65,700 → $82,280 +0.2% 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
+8.2% 21,400 → 23,100 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +8.2% 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.
~1,800 openings a year on average, including replacing people who leave.
WirerTest TechnicianInstrument TesterRepair TechnicianAvionics InstallerAircraft TechnicianAirplane TechnicianAircraft ElectricianAirplane ElectricianAviation ElectricianComposite TechnicianElectrical InstallerElectronic TechnicianInstrument SpecialistAutomatic Pilot MechanicAvionics Systems RepairerAircraft Armament MechanicMissile Facilities RepairerAircraft Instrument MechanicAirplane Electrical RepairerElectrical Aircraft MechanicMobile Electronics InstallerElectrical and Radio MechanicAviation Electrical Technician
Holding it up: embodiment . Weakest point: trust premium .
Tasks largely resist digitisation At 15, the resistant core is physical fault isolation — pin-to-pin continuity checks on a chafed harness behind a cockpit panel, tracing an intermittent that only appears at altitude or after a hard landing — while the parts AI genuinely takes (BITE code lookup, IPC and wiring-manual search, work-card drafting) keep it out of the 17-plus range where nothing offloads.
Hands-on in uncontrolled environments 18 reflects work done inside the airframe and on open ramps: contorting into avionics bays and wheel wells, crimping and pinning connectors by feel, hauling test sets and pitot-static rigs to the aircraft in whatever weather the gate has, with only bench work on removed LRUs happening in a controlled shop.
Licensed human required and personally liable 14 rests on the FAA certificate — A&P under 14 CFR Part 65 or repairman/repair-station authorisation under Part 145 — where your signature on the maintenance record and return-to-service entry is enforceable against you personally, short of the 18-20 band because much avionics work is signed off under a repair station's certificate rather than your own individual ticket.
Exists to be accountable for ambiguous calls 15 is the no-fault-found squawk: with no code, a schedule waiting and an MEL that may or may not cover it, you decide whether to release the aircraft or ground it, and that call is yours in the record rather than a step in a task card.
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 (15/20 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 (14/20) is whether the law requires a licensed human to sign. Trust premium (9/20) is whether buyers specifically pay for a person. Judgment and accountability (15/20) 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 38 of this occupation's 71 points (54%).
Embodiment (18/20) 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 81/100, still SAFE.
Task-mix shift is genuine here: once fault-code interpretation, IPC/wiring-diagram search and logbook drafting are absorbed, the residual day is intermittent-fault chasing, wire-bundle chafe hunting in bulkheads, and connector-level repair — the tier where no schematic predicts the fault. The score rises by subtraction of the routine tier rather than by any new capability barrier.
FAA formalizing that AI-generated diagnostic or maintenance recommendations cannot substitute for a certificated A&P/repairman's independent determination — e.g., an FAA policy statement or Part 43 rulemaking requiring the signing technician to document independent verification of any software-proposed corrective action, mirroring the agency's existing insistence that the RTS signature is personal and non-delegable. Similarly, an EASA Part-145 AMC addition on AI decision-support would propagate to US repair stations serving EU-registered fleets.
Avionics-specific certification tightening as fly-by-wire and NextGen ADS-B/CPDLC systems proliferate: an FAA requirement that installation and RTS of software-loadable avionics LRUs be signed by a repairman with type-specific authorization, narrowing who can sign at all. Data-load and software-config errors are already a recurring NTSB/SDR theme.
If the ground-or-release call becomes more consequential and more attributable — e.g., FAA SMS rules under Part 5 being extended to Part 145 repair stations requiring named accountability for deferral decisions and MEL applications, plus airline SMS audit trails naming the deciding technician — the exposure attached to the judgment increases.
Little headroom; already 18. The only route is fleet-level: composite airframes and dense integrated modular avionics push more work into confined-access inspection and bonding/grounding work that is harder to fixture for robots than legacy aluminum harness runs.
The limit. Trust premium has no realistic route: buyers are airlines and repair stations purchasing airworthiness compliance, not a human touch, and no passenger selects a carrier by who signed the RTS. Any perceived premium is the liability shield wearing a different hat. Total realistic headroom is roughly 8-10 points, nearly all of it regulatory, and it is offset over time by the automation of the paperwork tier and by OEM prognostic health-monitoring systems that shift diagnosis upstream to the manufacturer.
| Seattle-Tacoma-Bellevue, WA | 2,880 | $100,200 +22% |
| Dallas-Fort Worth-Arlington, TX | 870 | $87,090 +6% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 750 | $85,870 +4% |
| Atlanta-Sandy Springs-Roswell, GA | 650 | $89,180 +8% |
| Los Angeles-Long Beach-Anaheim, CA | 500 | $98,020 +19% |
| Phoenix-Mesa-Chandler, AZ | 340 | $78,690 -4% |
| Jacksonville, FL | 330 | $81,760 -1% |
| Oklahoma City, OK | 310 | $68,140 -17% |
| San Francisco-Oakland-Fremont, CA | 80 | $129,320 +57% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 80 | $117,470 +43% |
| Tulsa, OK | 120 | $108,840 +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 71. 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.