Industrial Decarbonization
AI Mapped Across the Construction Lifecycle: A Decade of Evidence
A PRISMA-based review of 10 years of AI research in construction, published in KeAi's Risk Sciences, grades maturity from predictive maintenance to digital twins and warns that compute footprints must offset claimed climate gains.

Waypoints
Systematic review covering 10 years of AI research in construction, published in KeAi's Risk Sciences journal
Research team drawn from Australia, Finland, Morocco and Thailand; corresponding author Fahim Ullah, Victoria University
Three principal pathways identified: cleaner and more resource-efficient energy use, decarbonization of the built environment, responsible production and consumption of resources
Relatively mature applications: predictive maintenance, building energy optimization, computer-vision monitoring; generative climate-resilient design and fully integrated digital twins remain at pilot stage
Six adoption barriers: poor data quality, interoperability, high implementation costs, organizational readiness, regulatory uncertainty, cybersecurity
Ten years of research on artificial intelligence in building and construction have been screened, coded and mapped in a systematic review published in KeAi's Risk Sciences journal, tracing AI applications from raw material production through to demolition.
The review was assembled by an international team of researchers in Australia, Finland, Morocco and Thailand. Following PRISMA guidelines, it pairs bibliometric analysis with content analysis to establish where AI is applied, how mature each application is, and what blocks adoption across a construction project.
The stakes are measured in tonnes and emissions. Buildings and construction sit among the most resource-intensive activities on the planet, consuming vast quantities of concrete, steel and aluminum and accounting for a substantial share of global energy use and energy-related carbon emissions. As extreme weather events intensify, the sector faces pressure both to reduce its climate impact and to become more resilient to it.
Fahim Ullah, corresponding author at Victoria University in Australia, frames the result as a systems argument rather than a tooling argument.
"AI should not be treated as an isolated technological application," Ullah said. "It is a lifecycle-wide enabling technology. It can support predictive decision-making, optimize energy and carbon, improve resource efficiency, automate processes and monitoring, and it works best when integrated with complementary tools such as building information modeling, digital twins and the Internet of Things."
Where AI enters the project lifecycle
The review locates AI across every stage of a construction project:
- Lower-carbon and more resilient material production
- Greener procurement and supply chains
- Energy-efficient and climate-resilient design
- Optimized equipment and scheduling on site
- Smarter building operation and maintenance
- Better recycling and waste recovery at end of life
From those application points, the team distilled a conceptual framework built on three principal pathways: cleaner and more resource-efficient energy use; decarbonization of the built environment; and responsible production and consumption of resources.
The evidence tilts toward mitigation
The findings are unevenly distributed. Most studies examine mitigation, such as cutting emissions and energy use, while far fewer address adaptation, such as climate-resilient infrastructure or extreme-weather preparedness. The maturity ranking splits in the same direction:
- Relatively mature: predictive maintenance, building energy optimization, computer-vision monitoring
- Largely at pilot stage: generative climate-resilient design, fully integrated digital twins
Six barriers recur across the literature, and no single fix clears them:
- Poor data quality
- Interoperability problems
- High implementation costs
- Organizational readiness
- Regulatory uncertainty
- Cybersecurity concerns
The authors also reject the assumption that computing capacity pays for itself in carbon terms.
"The presence of AI capability does not automatically produce a net-positive climate outcome," Ullah said. "Training and operating these models requires computational resources and electricity. The climate benefits of AI must be evaluated against its own environmental footprint."
What does this change for concrete, steel and demolition waste?
The lifecycle framing moves the value of AI in the materials sector out of the design office and into the recovery stream. For anyone trading recovered concrete, steel and aluminum, the review lists better recycling and waste recovery at end of life as a genuine AI-enabled contribution rather than a downstream afterthought. But the maturity ranking qualifies that claim: the review names predictive maintenance, building energy optimization and computer-vision monitoring as the comparatively mature applications, and it does not single out end-of-life recovery among them.
For purchasers, the operational consequence is that near-term gains in construction supply chains run through monitoring, scheduling and maintenance rather than through algorithmic redesign of material specifications, since generative climate-resilient design and fully integrated digital twins remain pilots. The barriers explain why. Interoperability between building information models, digital twins and Internet of Things sensors determines whether a contractor can prove a recovery rate; poor data quality determines whether anyone can verify it; and cybersecurity concerns determine whether facility and site operators will connect those systems at all.
What decides what happens next?
Regulatory uncertainty appears in the review's barrier list, and it is the item most directly within reach of procurement rules and standards. Until regulators settle how AI's own electricity and compute footprint gets counted against the carbon and material savings it produces, the net-positive claim stays unverifiable, and buyers cannot credibly compare a digitally optimized building against one built without it. The framework offers the checklist for that reckoning: three pathways, a maturity scale from computer-vision monitoring to digital twins, and six barriers that have to be answered in sequence rather than in isolation.
Ullah set the limit on the optimism. AI can make a major contribution to a low-carbon, climate-resilient construction industry, he said, but it is neither a standalone solution nor inherently sustainable.
via azocleantech.com (Source)
More from Grace Kim
Show full bio
Market editor covering money, funding and market shifts around Circular economy and the environmental industry at Circular Wire.
578 articles