EXPOSED
Roughly half this job is screen work AI already does competently — CAD detailing, tolerance and stress calculations, test-data reduction, spec comparison, and writing up test reports. The other half is genuinely physical: building and instrumenting prototypes, wiring strain gauges and thermocouples, running dynamometer and vibration rigs, and troubleshooting why a fixture is reading wrong. There is no license or personal sign-off protecting the role — the PE stamp sits with the engineer, not the technician — so the drafting-heavy tier shrinks while the lab-and-prototype tier holds.
Part 2020 shock, part continued decline in the years since.
Median pay $56,980 → $74,510 +4.6% 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
0% 38,300 → 38,300 on the projections basis
Exposed, and shrinking
Both signals point the same way: the tasks are largely automatable and the BLS projects 0% by 2034. This is the case where the score and the forecast agree, and it is the one worth taking seriously.
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.
~3,200 openings a year on average, including replacing people who leave.
Motor TesterTool AnalystBrinell TesterProcess AnalystSystems AnalystEngineering AideLaser TechnicianProof TechnicianTechnical AnalystOperations AnalystProcess TechnicianEngineering AnalystMechanical DesignerResearch TechnicianTool Design CheckerGyroscope TechnicianHydraulic TechnicianMechanical InspectorDurability TechnicianDevelopment TechnicianPerformance TechnicianExperimental TechnicianMechanical TechnologistEngineering Data Analyst
Holding it up: embodiment . Weakest point: liability shield .
Mixed — a routine tier and a judgment tier Generating a detailed drawing from a designer's layout, running a hand-calc stress check, reducing a run of thermocouple data into a plot, and drafting the test report are all now first-pass machine outputs, which drags the score to 9; what keeps it out of the 0-6 band is that setting up a fixture, deciding a load cell is reading badly, and re-rigging a test that failed for physical reasons remain hand-and-eye work no model performs.
Hands-on in uncontrolled environments You are in the lab bonding strain gauges, torquing bolts on a test article, hooking up a dyno, chasing a vacuum or coolant leak, and standing next to a spinning rotor — 14 rather than 18 because the environment is usually your own test cell with controlled access, not a customer's roof or an active production line.
No licence, no signature requirement Nothing you do requires a state license, and the drawing you detail and the test data you deliver both go out under the PE's or lead engineer's name — the 3 rather than 0 reflects that ASNT NDT levels, ASME certs, or an ABET-accredited AS are sometimes hiring filters, not legal gatekeeping.
Meaningful discretion Real calls exist — whether a suspicious data channel is instrument error or a genuine failure, whether to abort a run before you destroy the article — but they happen inside a written test procedure and get escalated to the responsible engineer, which is why this sits at 8 rather than in the teens.
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 (9/20 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 (3/20) 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 (8/20) 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 17 of this occupation's 40 points (43%).
Embodiment (14/20) 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.
Avionics Technicians SAFE
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 54/100, still EXPOSED.
Genuine two-tier structure: if CAD detailing, tolerance stack-ups and test-report writing are fully absorbed by generative CAD/LLM tooling (Siemens NX/Autodesk copilots already do drawing cleanup and report drafting), the residual job is instrumentation design, fixture debugging and deciding whether a suspect data channel is a real failure or a sensor artifact — work with no clean digital ground truth. The score rises because the automatable half stops being counted as part of the job, not because AI got worse.
Personal certification regimes that attach to the technician rather than the PE: ASNT NDT Level II/III certification, AWS CWI, and ISO/IEC 17025 accreditation clauses (A2LA, ANAB) that require a named, competency-authorized individual to approve test records and endorse method deviations. NQA-1 nuclear QA and NADCAP aerospace audits already require identified qualified personnel signatures on test data. If accrediting bodies extend those signature requirements explicitly to AI-generated data reduction — i.e. a named certified technician must attest the machine-reduced results — this moves from procedural to personal.
Growth in hardware-heavy programs that cannot be simulated to sign-off: DoD/aero qualification testing, battery and e-drive durability labs, hydrogen and heat-pump rig testing. Each adds strain-gauge/thermocouple installation, dyno and shaker-table setup and rig troubleshooting in one-off configurations. Watch for CHIPS/IIJA-funded test-lab buildouts and OEM battery test-cell expansions as the concrete signal.
Formal ownership of data-integrity and anomaly disposition: FDA 21 CFR Part 11 audit-trail rules for medical-device test data, and 10 CFR 21 defect-reporting duties, place a named individual on the hook for whether an out-of-spec reading is reported or dismissed. If AI-generated reductions require a human to record the disposition rationale under those regimes, the technician owns a consequential ambiguous call.
The limit. No realistic route to a trust premium — the buyer of a test report is an internal engineering group or a certifying body, neither of which pays extra for human-produced data, and the technician is not customer-facing. Liability also has a hard ceiling: as long as the PE stamp sits with the engineer, technician certifications shield the task, not the person's exposure to being replaced by a cheaper technician plus better tooling.
| Detroit-Warren-Dearborn, MI | 3,140 | $69,490 -7% |
| Los Angeles-Long Beach-Anaheim, CA | 1,220 | $79,080 +6% |
| Boston-Cambridge-Newton, MA-NH | 1,090 | $62,390 -16% |
| Houston-Pasadena-The Woodlands, TX | 1,030 | $91,450 +23% |
| San Francisco-Oakland-Fremont, CA | 790 | $100,620 +35% |
| Kansas City, MO-KS | 600 | $100,150 +34% |
| New York-Newark-Jersey City, NY-NJ | 540 | $73,400 -1% |
| San Jose-Sunnyvale-Santa Clara, CA | 530 | $96,680 +30% |
| Kennewick-Richland, WA | 90 | $126,230 +69% |
| Baton Rouge, LA | 90 | $119,630 +61% |
| Bakersfield-Delano, CA | 90 | $110,030 +48% |
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 40. 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.