Construction and Real Estate and Artificial Intelligence: The Slowest Adopter, Thin Evidence, and Two Applications Where the Results Are Already Visible

Abstract

Purpose. Construction is a large industry with a reputation for slow technological change, and real estate is a market in which valuation is a prediction problem. This article asks where artificial intelligence stands in construction and real estate in 2026 and where they may be by about 2035. Design and methods. A review combining the share of the construction and real estate research literature that involves AI (OpenAlex, 2010–2025), Eurostat's survey of AI use in EU construction enterprises (2021–2025), a literature review of AI in construction, a review of field demonstrations of AI control of building heating and cooling, the well-documented failure of an automated home-buying business, and the US rule on automated valuation models. Claims taken from the literature were checked against the sources cited. The scenarios are judgements built on this evidence. Findings. The share of construction and real estate works that mention AI rose from 1.8% in 2018 to 11.9% in 2025, 6.6 times as much. Use lags: in the EU, 10.8% of construction enterprises used AI in 2025, the lowest of the twelve sectors with Eurostat data and about half the 19.9% for all enterprises, after a dip from 4.6% in 2021 to 3.2% in 2023, and 88.2% used none. Only 3.3% used three or more technologies. We found few rigorous studies of what AI does to productivity on construction projects; the literature is largely reviews, surveys of perceptions and case studies. The clearest evidence is in building operations and in valuation. A review of 104 field demonstrations of predictive and reinforcement-learning control of heating and cooling found that 71% used protocols that may give unreliable estimates, and that among the rest the cost savings were about 13% in commercial buildings. In real estate, a large online platform closed its home-buying business in 2021 after an inventory write-down of $304 million, having been unable to forecast home prices accurately enough, and US regulators now require quality-control standards for automated valuation models, in force since October 2025. We expect AI to enter construction first through documents, scheduling and site monitoring, and building operations to be the largest measurable gain by 2028; whether construction productivity changes in 2029–2032 depends on the industry's structure more than on the technology. Originality. The article identifies the sector with the lowest measured AI use in the EU, states plainly that rigorous evidence of productivity effects is thin, and sets two documented applications, one a measured gain and one a costly failure, against that background.