SAFE
The work is hoistway rigging, aligning rails and guide shoes, pulling and terminating traveling cables, swapping controller boards, adjusting door operators and brakes, and troubleshooting intermittent faults in decades-old equipment inside cramped pits and machine rooms — none of which a language model or any shipping robot performs. AI genuinely helps with fault-code lookup, wiring-diagram search, service-report writing, and predictive maintenance scheduling from remote monitoring data, which trims paperwork and diagnostic time rather than headcount. Most states license elevator mechanics, inspections must be signed by a certified human, and the life-safety liability plus strong IUEC representation make substitution both illegal and impractical; note licensure is a regulatory shield that legislatures could narrow.
Nearly all of this fall was the 2020 shock. It has been climbing back since.
Median pay $84,990 → $109,910 +3.5% 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
+5% 24,200 → 25,400 on the projections basis
Hard to automate, and growing
The work resists current AI and the BLS projects +5% 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.
~2,000 openings a year on average, including replacing people who leave.
InstallerElevator WorkerElevator BuilderElevator ErectorElevator AdjusterElevator ExaminerElevator MechanicElevator RepairerElevator InstallerEscalator MechanicBuilding ServicemanContract ServicemanElevator ServicemanEscalator InstallerElevator ConstructorElevator TroubleshooterFreight Elevator ErectorElevator Service MechanicRepair Maintenance WorkerEscalator Service MechanicElevator Installation WorkerHydraulic Elevator ConstructorEscalator Maintenance TechnicianElevator Technician (Elevator Tech)
Holding it up: embodiment . Weakest point: trust premium .
Tasks largely resist digitisation A 17 rather than a 20 reflects that remote monitoring and fault-code diagnostics genuinely absorb part of the troubleshooting call — some no-fault-found trips get resolved from a laptop — but shimming rails to a thousandth over a 40-story hoistway, re-roping a traction machine, and re-timing a door operator against a worn cam remain entirely manual.
Hands-on in uncontrolled environments Full 20: the job is performed on top of a moving car, in a flooded pit, on a hoistway ladder, and in an unventilated machine room, with counterweights, live 480V mainline disconnects and open shaft edges — conditions no fielded robot can even be inserted into, let alone work in.
Licensed human required and personally liable At 15, most states and major cities license elevator mechanics under ASME A17.1-adopting codes and require a named certified person to sign off Category 1 and Category 5 tests, so the mechanic's card is personally on the line for a life-safety device — short of the 18-20 band only because a handful of states still have no mechanic licensure and helpers work under someone else's ticket.
Exists to be accountable for ambiguous calls A 16 is earned every time you decide whether an intermittent leveling fault means run it another week or lock the car out of service, whether a corroded governor rope passes or gets condemned, and how to safely handle an entrapment before the fire department arrives — calls made alone, off-code-judgment, with a fatality as the downside.
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 (17/20 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 (15/20) 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 (16/20) 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 42 of this occupation's 79 points (53%).
Embodiment (20/20) 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 88/100, still SAFE.
Task-mix shift as OEM remote diagnostics (Otis ONE, KONE 24/7, Schindler Ahead) absorb the routine callback triage: what remains on the truck is intermittent-fault chasing on 40-year-old relay logic and non-OEM controllers no monitoring platform covers, plus modernization work — the judgment tier grows as a share of the day
Remaining unlicensed states (e.g., several southern/mountain states without elevator mechanic licensure) adopting the ASME A17.1/QEI-based licensing model already pushed by NAESA and IUEC, plus state rules requiring a licensed mechanic to personally sign off on any remote-monitoring-triggered or AI-recommended adjustment before the car returns to service — making the human the named liable party for predictive-maintenance decisions rather than the OEM's monitoring platform
Post-incident code change after a high-profile entrapment or fall: ASME A17.1 committee adding a requirement that door-restrictor and brake adjustments be witnessed and countersigned by a second certified mechanic, as some jurisdictions already require for full-load safety tests
Wider adoption of contract language and state rules making the responding mechanic the person who decides to take a unit out of service (red-tag authority) rather than a building manager or dispatch algorithm — already the norm in some jurisdictions and a live IUEC bargaining item where OEM remote centers try to override field calls
The limit. Already 79/100 with embodiment maxed and task_resistance near ceiling; realistic headroom is a few points in liability_shield and judgment_accountability. Trust_premium has no plausible route — building owners buy from OEM/service contractors on price and uptime, not from named humans, and no consumer-facing preference for a human mechanic exists. The genuine downside risk is the reverse: state legislatures narrowing licensure scope or allowing remote inspection sign-off would cut liability_shield faster than any of these levers add.
| New York-Newark-Jersey City, NY-NJ | 3,270 | $136,000 +24% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 1,050 | $108,170 -2% |
| Chicago-Naperville-Elgin, IL-IN | 850 | $146,580 +33% |
| Los Angeles-Long Beach-Anaheim, CA | 850 | $140,600 +28% |
| Houston-Pasadena-The Woodlands, TX | 540 | $104,350 -5% |
| Philadelphia-Camden-Wilmington, PA-NJ-DE-MD | 540 | $132,640 +21% |
| Atlanta-Sandy Springs-Roswell, GA | 510 | $83,500 -24% |
| Dallas-Fort Worth-Arlington, TX | 500 | $108,150 -2% |
| San Jose-Sunnyvale-Santa Clara, CA | 70 | $173,920 +58% |
| San Francisco-Oakland-Fremont, CA | 370 | $168,270 +53% |
| Chicago-Naperville-Elgin, IL-IN | 850 | $146,580 +33% |
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 79. 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.