Artificial intelligence’s impact on work and workers is attracting attention—and concern—across the country. Both younger and more advanced workers are worried about AI’s impact on current and future jobs, including the need to upskill, reskill, or even consider new career paths entirely. Employers in the technology sector as well as other industries are not only driving increased usage of AI in many cases, but are also adapting to it in real time, with the potential for widespread disruption and innovation. And policymakers, training providers, and other workforce and education leaders are simultaneously scrambling to respond and preparing for what comes next.
Amid all this change and uncertainty, one point is clear: the need to equip workers with skills and experience that provide greater flexibility and opportunity for long-term career growth. Positions in the building trades—and across the built environment more generally—offer just that.
Past Brookings research on the “infrastructure workforce”—workers directly involved in constructing, operating, and maintaining the country’s transportation, water, energy, broadband, and other physical assets—has revealed an enormous segment of the labor market with growing hiring needs, generally lower formal educational barriers to entry, and more competitive and equitable pay. These positions, moreover, emphasize work-based learning opportunities (e.g., apprenticeships and internships) that promote portable and transferable skills over time. In other words, whether it’s electricians and plumbers or engineering technicians and construction managers, these positions intuitively appear well positioned to offer greater certainty—or “durability”—in the face of AI disruptions.
New Brookings research expands on this earlier infrastructure workforce analysis to consider the broader “built environment workforce”—a collection of 148 occupations for which we have complete data—to illustrate how many roles involved in construction, design, and engineering are “AI-durable,” or less exposed to AI. In particular, we aimed to define a clear and consistent set of positions across the built environment to better understand their ability to offer more AI-resistant and/or AI-adaptable career pathways for workers, using two primary measures:
- AI exposure: The AI Occupational Exposure (AIOE) score developed by Felten, Raj, and Seamans (2021) gives us a standardized way to compare exposure across specific built environment occupations. Occupations with above-average scores (greater than zero) are more exposed to AI, while occupations with below-average scores (less than zero) are less exposed.
- AI complementarity: Different studies find that physical, manual, and craft occupations sit at the low end of AI exposure, but leave open whether AI will work with or instead of workers in a given role. Pizzinelli et al. (2023) address this by augmenting the AIOE score with a complementarity index (or “Theta score”), allowing us to gauge the ways in which AI is complementing work in this space. Occupations with above-average scores (greater than 0.5725) have higher AI complementarity, while occupations with below-average scores (less than 0.5725) have lower AI complementarity.
The rapidly evolving nature of AI and how we measure it is a complex exercise garnering much deserved attention across the research community, and warrants deeper exploration beyond this piece. In addition, the infusion of AI into robotics and other industrial tools implies almost certain inroads into production and built environment activity in the future. But by looking at AI exposure and complementarity across the built environment, this preliminary analysis reveals several promising characteristics of these positions to support greater economic opportunity.
The vast majority of the built environment workforce is less exposed to AI compared to the rest of the US workforce
The 148 built environment occupations we analyzed directly employ 17.3 million workers in the construction, design, and engineering of different projects across the country. Of these workers, we found the vast majority (83.6%, or 14.5 million workers) are employed in occupations with less AI exposure as measured by the AIOE score. These include large building trades positions such as maintenance and repair workers (1.7 million workers), construction laborers (1.1 million), and electricians (772,000), but also several smaller, more specialized “green” positions such as solar installers and forest and conservation workers. As Table 1 shows, the 10 largest occupations with below-average AI exposure account for almost half of total built environment employment.
The remaining 33 built environment occupations more exposed to AI include a collection of engineering and architectural roles, such as civil engineers, landscape architects, and urban and regional planners. This higher exposure concentrated in desk-based and other “white collar” occupations is consistent with concerns found across the broader labor market.
AI exposure is concentrated in a smaller subset of higher-paying built environment occupations
While the 148 built environment occupations we analyzed pay higher median wages ($65,011 annually) overall compared to all occupations nationally ($61,672), there were notable differences depending on the level of AI exposure.
For instance, the median annual wage of the 115 occupations less exposed to AI is $58,153—a figure influenced heavily by large occupations such as carpenters and welders. In contrast, the median annual wage of the 33 occupations more exposed to AI is $100,105; these positions include construction managers, geoscientists, and other higher-paying managerial and engineering roles. This bifurcation shows that exposure is most concentrated in a smaller set of higher-paying built environment roles.
Three-quarters of the built environment workforce has higher AI complementarity compared to the entire US workforce
While most built environment workers face less AI exposure, that does not mean they will never use AI at work. In fact, 73.8% of these workers—or 13 million of the 17.3 million—have higher AI complementarity based on the Theta score. Put another way, most built environment workers—from electricians and pipelayers to heating, air conditioning, and refrigeration mechanics—who use AI in their jobs will find it to be a complement rather than a substitute for their labor.
The remaining 63 built environment occupations that have less AI complementarity are concentrated in relatively smaller roles, employing 4.5 million workers total. These include an assortment of construction roles, such as tapers and tile and stone setters, as well as occupations involved in producing physical inputs for projects, such as machine feeders and etchers and engravers.
Wages tend to be higher in built environment occupations with greater AI complementarity
Familiarity and use of AI in this work tend to result in higher wages: The median annual wage is $69,491 for the 85 built environment occupations with greater AI complementarity. These include many managerial and supervisory roles in engineering and construction, in addition to a range of other large occupations such as civil engineers and architects as well as structural iron workers and telecommunication line installers.
By comparison, the median annual wage is only $52,400 for the remaining 63 built environment occupations, again including many occupations involved in producing material inputs.
Broadening research into the AI durability of the built environment workforce
Understanding AI’s impact on the built environment workforce is not a one-time exercise, but an ongoing one given how rapidly both AI capabilities and the measurement tools we use to track them are evolving. The preliminary evidence presented here suggests that the built environment is a particularly worthwhile focus for continued research and investment. These occupations are essential to the country’s physical infrastructure—and to the construction of the data centers and energy systems that the digital economy, including AI itself, depends on.
Analysis of how education levels, demographic characteristics, and geographic context shape AI durability would sharpen this picture considerably. Future research should continue to examine these dynamics, and consider how career pathways, training systems, and workforce development investments can best align with the durable strengths this sector demonstrates.
Authors
The Brookings Institution is committed to quality, independence, and impact.
We are supported by a diverse array of funders. In line with our values and policies, each Brookings publication represents the sole views of its author(s).
