EXPOSED
Almost nothing a carpenter's helper does — carrying lumber, holding forms in place, sweeping the deck, pulling nails, setting up scaffolding and cutting stock to marked lines — can be done by a language model, so direct AI exposure is near zero. The real pressure is different: no license, no client relationship, no ownership of decisions, and rising factory prefabrication and panelized framing that move framing labor off the jobsite into controlled shops where automation actually works. This is a stepping-stone title, and its safety depends entirely on whether the helper converts to a journey-level carpenter.
Part 2020 shock, part continued decline in the years since.
Median pay $33,060 → $43,780 +5.9% 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
+4.5%
Percentage only. The projection counts a different population from the 21,680 above — it includes self-employed workers, which for this occupation is most of them, so the two headcounts are not comparable.
Exposed, but growing
AI can already do a lot of these tasks, and the BLS still expects +4.5% more of these jobs by 2034. Demand for the output is growing faster than the work is being automated away — the mechanism BLS gives for software developers, and the combination people most often misread as an error.
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.
~2,700 openings a year on average, including replacing people who leave.
HelperCooper HelperJoiner HelperCarpenter MateHammerer HelperCarpenter HelperShipwright HelperCarpenter's HelperBeam Builder HelperCabinetmaker HelperCarpenter AssistantTimber Framer HelperHouse Carpenter HelperBuilding Carpenter HelperConstruction Carpenter's HelperHardwood Floor Installation Helper
Holding it up: embodiment . Weakest point: liability shield .
Tasks largely resist digitisation Hauling 12-foot studs up a stair tower, bracing a wall while the carpenter shoots plates, stripping snap ties off green concrete forms and hand-digging around a footing are irreducibly physical, sequence-dependent tasks that no software touches — the 16 rather than 19 reflects that some of the cut list, layout dimensions and material takeoff you work from is already generated upstream, and prefab panels arriving pre-assembled delete the very tasks you'd otherwise be doing.
Hands-on in uncontrolled environments Your workday is outdoors on uneven mud and rebar in whatever weather the schedule ignores, moving between ladders, scaffold planks and open floor edges with loads in both hands — an 18 rather than 20 only because a share of helper hours happen in shops, garages or enclosed remodels rather than open steel and dirt.
No licence, no signature requirement Nothing in this title requires a license or certificate; OSHA 10 or 30 might be a site condition and fall-protection or scaffold-user training gets logged, but the permit is pulled in the contractor's name, the inspector signs off on the journeyman's work, and no code or statute makes your signature necessary to anything — hence the 1 rather than a clean 0.
Executes defined procedures on defined inputs You cut to marks someone else made, stack material where the lead tells you, and set scaffold per the manufacturer's frame configuration — the small discretion you do own is real safety judgment (refusing an unsafe lift, calling out an unbraced wall, spotting a missing plank), and that's what puts this at 4 instead of 0.
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 (1/20) is whether the law requires a licensed human to sign. Trust premium (4/20) is whether buyers specifically pay for a person. Judgment and accountability (4/20) 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 9 of this occupation's 43 points (21%).
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.
Roofers 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 53/100, still EXPOSED.
Registered-apprenticeship mandates on subsidized work: the IRA clean-energy prevailing-wage/apprenticeship provisions (Treasury final rule, 2024) require a labor-hour percentage performed by registered apprentices and journeyworker-to-apprentice ratios. If similar apprentice-ratio conditions spread to state infrastructure and affordable-housing funding, the helper slot becomes a contractually required headcount on the job rather than a discretionary one.
Task-mix shift toward field retrofit and remediation: if panelized/prefab absorbs new-build repetitive framing, the residual on-site work concentrates in cut-and-fit against out-of-plumb existing structures, panel setting, seam closure and punch-list correction — the tier prefab shops cannot pre-cut. This raises resistance for the surviving on-site roles while shrinking their count; the score moves but the headcount does not.
Expansion of certification-gated helper tasks: OSHA already requires a qualified/competent person for scaffold erection (1926.451) and a certified signalperson for crane picks (1926.1428). If state boards or GC insurers push further — e.g. mandatory card-carrying certification for fall-protection anchor installation, shoring/formwork bracing, or powder-actuated tools, with the cardholder named on the daily log — parts of helper work become person-attached rather than body-attached. Watch state OSHA plans (WA, CA, OR) and GC prequalification checklists.
Formal stop-work authority written into contract: some union agreements and large-GC safety programs (and CPWR-promoted safety-climate language) give any worker documented authority to halt an activity for imminent hazard, with the call logged. Where that authority is contractual and the log is auditable, the helper owns a consequential call rather than only executing directions.
The limit. Embodiment at 18 is effectively maxed and is the only strong score; the realistic ceiling is roughly the low 50s. Trust premium has no plausible route — buyers hire the contractor, never the helper, and no client will ever pay a premium for a specific human carrying lumber. The dominant risk is not AI doing this work but prefabrication relocating it into shops with fewer helpers, and none of the levers above reverse that.
| New York-Newark-Jersey City, NY-NJ | 1,360 | $48,850 +12% |
| San Juan-Bayamon-Caguas, PR | 1,010 | $26,380 -40% |
| Los Angeles-Long Beach-Anaheim, CA | 750 | $64,810 +48% |
| Houston-Pasadena-The Woodlands, TX | 740 | $44,600 +2% |
| Washington-Arlington-Alexandria, DC-VA-MD-WV | 630 | $44,380 +1% |
| Dallas-Fort Worth-Arlington, TX | 530 | $43,740 +0% |
| San Diego-Chula Vista-Carlsbad, CA | 500 | $45,710 +4% |
| Miami-Fort Lauderdale-West Palm Beach, FL | 470 | $40,110 -8% |
| Los Angeles-Long Beach-Anaheim, CA | 750 | $64,810 +48% |
| San Francisco-Oakland-Fremont, CA | 340 | $62,370 +42% |
| Santa Cruz-Watsonville, CA | 50 | $59,090 +35% |
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 43. 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.