Construction workers in Phoenix now have an invisible safety net powered by artificial intelligence. Researchers at Arizona State University have deployed a heat stress early-warning system that analyzes real-time data from wearable biosensors and environmental monitors. When the system detects early signs of fatigue or heat stress, it sends personalized alerts before the worker feels symptoms. The technology, developed by the Safety Automation and Visualization Environment Laboratory (SAVE Lab) with industry partners Achen-Gardner Construction, Willmeng Construction and CHASSE Building Team, aims to prevent heat-related illness on jobsites where summer temperatures routinely top 110 degrees.
"Each of these is difficult to observe until it becomes dangerous, and that is exactly where AI helps," said Siyuan Song, an associate professor of construction engineering who leads the SAVE Lab at the Ira A. Fulton Schools of Engineering. "It monitors conditions no human observer could track, recognizes that two construction workers in identical conditions can respond differently, and shifts safety management from reacting to incidents toward preventing them."
AI moves safety from reactive to preventive
Song's lab creates immersive virtual-reality safety training environments to study how workers learn and respond to hazards. Her team also uses large language models to analyze accident reports, study extreme weather impacts, and evaluate safety training programs. The core insight driving this work: physical strain, psychological stress, and heat stress are hard to spot until a crisis hits. AI catches patterns invisible to even experienced supervisors.
Shiva Pooladvand, an assistant professor of construction engineering, leads the Pooladvand Research Group, which develops human-centered construction technologies at the intersection of robotics, computer science, cognitive science, and virtual reality. "I use AI to analyze multimodal data captured from workers and the environment to identify patterns related to workers' decision-making, behavior and safety to anticipate unsafe working conditions or identify workers who may be at risk," Pooladvand said. "This enables more personalized interventions to enhance safety, automation and productivity."
Training the workforce with virtual reality and digital twins
At the Construction Workforce and Technology Lab, assistant professor Ricardo Eiris combines AI with human-computer interaction, virtual reality, digital twins, and drones. The goal is to expand how people learn about construction sites beyond the limits of physical classrooms or active jobsites. Eiris uses AI to model human behavior and create responsive systems that improve learner performance in construction settings.
Kenn Sullivan, a professor of construction management, applies AI to statistical modeling, visualization, and documenting research workflows. His approach is pragmatic: AI helps his team test analytical methods and pass institutional knowledge to the next student joining the group.
This focus on practical application surfaced at the recent AI for Construction 2026 Contractors Summit, hosted by the Del E. Webb School of Construction and the new AI in Construction Consortium. Industry leaders shared how they use AI in agent-based workflows for business development, virtual design, safety, visual analysis, scanning, and project controls. "The contractors summit and the proposed industry consortium originated from our executive council as opportunities for construction companies, technology providers, faculty members and students to learn from one another in a collaborative environment," said Tim Becker, programs chair of the school.
Preparing students for AI-augmented careers
Faculty members are redesigning how they teach as AI tools grow more capable. Sullivan observed that over nine months, AI went from completing almost none of his class assignments to finishing up to 60% of them. That forced a hard look at assessment. His AI in Construction course now requires students to analyze data, build custom AI skills and agent-based workflows, develop working apps, and create portfolio websites to showcase their solutions.
Pooladvand's students discuss real construction project applications of AI and learn to verify AI-generated information using engineering judgment. Song's Advanced Construction Safety Engineering and Management course uses the SAVE Lab's VR safety training and jobsite sensing deployments as case studies and hands-on experiences.
"In many ways, AI makes that foundational knowledge even more important," Sullivan said, referring to construction methods, estimating, scheduling, safety, contracts, and project management. Song put the stakes plainly: "Responsible AI use is essential in our field for a simple reason: In construction, decisions have physical consequences. An unverified AI output in an essay is an academic problem. An unverified AI output in a safety plan, cost estimate or schedule is a potential hazard to people and projects."
Why this matters for real estate and construction professionals
The ASU research signals a shift in how jobsites will manage risk. Wearable sensors paired with AI can catch heat stress and fatigue before a worker collapses, which has direct implications for liability, insurance costs, and OSHA compliance. For developers and general contractors, these tools offer a way to reduce recordable incidents and keep projects on schedule during extreme weather. The industry consortium model also points to a faster path for adoption: companies can test emerging safety technologies in collaboration with university labs rather than experimenting on live projects alone. As AI becomes embedded in training, estimating, and project controls, firms that train their teams to verify AI outputs with engineering judgment will hold an edge over those that treat the technology as a black box.
Your membership also unlocks:
