As generative artificial intelligence tools gain traction across construction workflows, experts are raising concerns about the risks posed by inaccurate outputs—particularly when those outputs appear authoritative.
According to Max Mahdi Roozbahani, a senior lecturer specializing in machine learning and natural language processing at Georgia Tech, the construction industry must remain cautious when integrating AI into critical decision-making processes.
Imagine a team takes a photo near a jobsite. In the background, a work truck on the highway flashes a red beacon.
A human understands the context: It is likely a utility vehicle. A generative artificial intelligence system might label the beacon as “police presence,” infer there was an accident and summarize the day as an incident. That one incorrect label can appear in a daily report, a safety log or a claim file months later, complete with a timestamp and a confident narrative that no one intended to create.
Construction firms are rapidly adopting generative AI copilots to search for and summarize project documents, emails and schedules. The goal is speed. With tight margins and chronic labor shortages, the industry is eager to improve efficiency.
However, experts caution that AI-generated outputs should not be treated as definitive or used for final approvals.
“I teach machine learning at Georgia Tech, and my background spans computer science and civil engineering across both academia and industry. That vantage point makes one risk apparent: Teams conflate well written answers with ground truth. In construction, ground truth is what is physically installed and supported by field evidence, including photos, timestamps and locations. But not text alone.”
In sectors such as finance or law, written documentation often defines reality. In construction, however, documentation can lag behind actual conditions or contain inaccuracies. Reports, submittals and invoices may be incomplete, outdated or incorrect.
This creates a major vulnerability when AI systems rely on such records. Even when summarizing accurately, the underlying data itself may be flawed.
“The danger is most significant in work that becomes invisible once covered: foundations, reinforcing steel, post-tensioning, fireproofing and critical mechanical, electrical and plumbing routing. These are the places where errors become catastrophic.”
Roozbahani highlights how AI can misinterpret evolving project data. For example, an outdated geotechnical report might be used to confirm compliance, even if later revisions introduced critical changes.
“If a generative AI assistant is asked, ‘Is the foundation plan compliant?’ it might retrieve an older revision, omit a conditional statement and reduce engineering nuance to a definitive ‘Yes.’”
“If that sentence influences a decision, the project drifts from engineering judgment toward automated optimism. The first time those assumptions are tested may be after a disaster.”
Despite these risks, AI remains a powerful tool when used appropriately.
“A reasonable objection is clear: If humans must verify every output, what is the point of AI?”
“The point is leverage. Many tasks in construction are repetitive: drafting routine communications, summarizing meetings, mapping submittals and flagging missing attachments. If AI reduces time spent on repetitive work, engineers can spend more time on field checks and verification.”
AI can streamline administrative tasks and improve productivity, allowing professionals to focus on validation and on-site oversight—areas where human judgment remains essential.
However, Roozbahani emphasizes that successful implementation requires training and process changes, not just technology adoption. Teams must understand data sources, system limitations and how to structure queries effectively.
“Prompting matters. Vague questions invite vague answers. Safer prompts require citations, list assumptions and force the system to say what it cannot confirm. These practices reduce hallucinations, but they do not eliminate them.”
To mitigate risks, industry leaders are encouraged to establish strict standards for AI use in compliance, safety and claims management. These include requiring traceable sources, timestamps and supporting documentation for every AI-generated output.
“Generative AI can make construction faster. It can help teams find information buried across folders and emails. But the built environment depends on proof, not prose. When chatbots hallucinate, infrastructure pays.”
The growing adoption of generative AI across construction reflects broader digital transformation trends in the industry. From document management to predictive analytics, firms are increasingly relying on data-driven tools to improve efficiency and reduce costs.
However, construction differs from many other industries due to its reliance on physical verification. Unlike purely digital sectors, errors in construction can have long-lasting consequences, including structural failures, safety incidents and legal disputes.
The concept of “AI hallucination”—where systems generate plausible but incorrect information—poses unique risks in this context. When such outputs are embedded into official records like safety logs or compliance reports, they can create liabilities that persist long after project completion.
As a result, experts stress the importance of combining AI capabilities with robust human oversight. Rather than replacing expertise, AI should function as a support tool that enhances decision-making while maintaining accountability.
Looking ahead, advancements in multimodal AI—systems that integrate visual and textual data—may improve accuracy. However, these technologies also introduce new complexities and potential failure points, reinforcing the need for cautious and informed implementation.
Originally reported by Max Mahdi Roozbahani in Construction Dive.
