Builders and developers are committing serious capital to artificial intelligence (AI) as they confront labor shortages, margin pressure and growing project complexity—but many are still determining where it’s most valuable, according to BRG’s new research report AI in Construction: Measurable Results, Unclear Benchmarks.
According to survey responses from more than 100 construction and development executives, AI adoption now spans the full project lifecycle: pre-construction (46%), construction (49%), post-construction (40%) and business operations (43%). Planning and feasibility is the most widely adopted individual function (58%), rising to 80% among builders. Nearly seven in 10 (68%) respondents have spent at least $250,000 on AI over the past 24 months, and 37% have invested $1 million or more. But just 15% say AI is fully integrated across projects and operations.
“Planning and feasibility is a natural starting point for builders because that is where much of the project risk sits,” said Dara Hendrix, a managing director at BRG. “It also involves some of the most time-consuming work—developing conceptual estimates and other early analysis—while often generating little or no direct revenue. AI can help teams make better-informed decisions earlier.”
Across most functions, respondents were more likely to cite gains in effectiveness—better quality, performance and outcomes—than in efficiency. Double-digit cost savings were widespread, reported during construction (67%), in business operations (62%) and across pre- and post-construction (58% each), with nearly 20% of respondents citing post-construction savings greater than 30%.
Workforce effects, meanwhile, are showing up in structure rather than headcount. Only 11% of respondents report reducing staff because of AI, while 18% say it has led them to add staff in certain areas. More than half have changed their reliance on outside vendors, consultants or contractors; 46% have changed team structure and collaboration; and 38% have reskilled or retrained existing employees.
Obstacles to AI adoption remain plentiful and formidable. Respondents most often cite a lack of skilled talent or expertise (40%), limited relevance to their work (38%) and data and integration challenges (37% each) as factors likely to limit AI’s benefits. Risk concerns start with liability for errors in AI-generated work (43%) and unclear treatment of AI in contracts (40%)—yet only 42% of executives say their policies address human oversight and accountability and just 34% say they address accuracy and reliability.
Measuring returns presents its own challenge, with 37% of respondents primarily relying on financial metrics, 29% on operational measures and about one-quarter on usage or adoption, while 6% do not evaluate ROI at all.
“The hardest ROI questions arise when AI makes something possible that was not realistically on the table before,” said Travis Olson, a director at BRG. “If the technology can review two million emails, the alternative may not be a slower review; it may be no review at all. If you focus too narrowly on ROI, you risk missing out on broader changes.”
