Construction AI exposure is low where the work is physical and high in exactly two places, according to an NAHB analysis of federal data published Oct. 1. Of 47 construction occupations the home builders’ economists pulled from a new Bureau of Labor Statistics table, 45 rate “low” or “moderate.” That’s about 96%. The two rated “high” are construction managers and construction and building inspectors. None rate “very high.”
Carpenters, laborers, roofers and operating engineers all sit in the low group. No surprise there. Software doesn’t set trusses.
What the BLS AI exposure categories measure
The BLS exposure categories are a new supplement to the agency’s 2025-35 employment projections. They blend five outside studies. Three are theoretical ratings of which tasks a language model could do. Two are observed usage data from commercial AI assistants, mapped back to occupations.
BLS is unusually blunt about the limits. A category isn’t a forecast of job growth or decline, isn’t a wage forecast and isn’t a worker replacement estimate. It’s relative, too: “high” means high compared with other occupations, not high in any absolute sense.
One caveat deserves more attention than it’ll get. BLS says the theoretical sources treat AI capabilities as those available no later than mid-2023. The models in use today read drawings and photos, which the 2023 versions largely couldn’t. An exposure score built on that vintage probably understates what’s coming for anyone whose job involves a plan set.
Construction managers and inspectors are the exceptions
NAHB’s explanation for the two outliers is that both jobs run on planning, documentation, scheduling, compliance and reporting. That’s text, and text is what these tools handle.
The projections attached to them point in different directions. BLS expects construction manager employment to grow 9% from 2025 to 2035. Inspector employment is projected to stay essentially flat. Same exposure category, very different decade, which is BLS’s own point about exposure not predicting headcount.
The fastest-growing occupation in NAHB’s group is solar photovoltaic installer, projected up 37% with a moderate rating.
There’s a scope limit worth knowing. The 47 occupations are the construction trades in the federal classification system plus construction managers. Estimators, engineers, architects, accountants and administrative staff aren’t in the count, though construction firms employ plenty of them and their work is the most document-heavy in the building.
AI in construction jobs: where the pressure lands first
So the headline number is reassuring and a little beside the point. The field is insulated. The trailer and the home office aren’t, and the home office was left out of the tally.
That lines up with what contractors report doing. NAHB says builders in its own survey use AI mostly for marketing, project planning and design. When Suffolk and MIT put numbers on construction AI last month, the gains they described were in cost and schedule management. And the share of contractors piloting automation nearly tripled in a year by one benchmark, with accuracy cited ahead of labor savings.
For a contractor, the practical read is about who gets retrained first. It’s the project engineer buried in submittals and the superintendent writing daily reports at 6 p.m., not the crew. BLS projects 9% more construction managers by 2035. What each of them does all day is the part the table can’t tell you.