Artificial intelligence (AI) has been a mainstay in construction tech conversations for the past few years, and it continues to be a hot topic today. As the fastest-evolving technology in the past four decades, we expect this to continue in the foreseeable future.
Research by Autodesk found that 78% of construction leaders believe that artificial intelligence will enhance our industry, and 66% say that AI will be essential across the board in two to three years.
But even the most promising innovations require thoughtful consideration. According to Autodesk’s 2025 Design & Make Report, only 40% of business leaders say they are approaching or achieving their AI goals. From security concerns to fears of job displacement, we must address many valid concerns to get the most out of AI without sacrificing ethics, quality, and security.
In a recent Autodesk-hosted panel, we brought together a great lineup of construction leaders to unpack the realities of AI adoption in the field and the office. They included:
These experts shared real-world insights into what's working, what's evolving, and where the industry is headed. Here are the top takeaways from the discussion.
Many companies are at different stages of their AI journey—some are just beginning to digitize their data, while others are already leveraging predictive analytics. Approaches to AI adoption vary widely depending on the organization's culture, needs, and goals.
At Fortis Construction, the team approaches AI with a mindset of curiosity and openness, rather than fear or distrust. Mitch said, "We're very much looking at it like we don't know what impact this could have, but we're going to seek it out." He references Amara's Law—the idea that we tend to overestimate the impact of technology in the short term and underestimate it in the long term—as a useful framework for thinking about AI adoption.
"I see that pretty aggressively with AI right now," Mitch said. "There's been a lot of hype, a lot of discussion, a lot of effort around how AI is going to change our world today. I don't think any of us have seen that, but I do expect it to have a profound impact in the future."
When it came to adopting AI within Hazen and Sawyer, Jamie shared that the company took a decentralized and inclusive approach. "We didn't create one committee to look at this,” Jamie added. “We have an inclusive approach because we have different types of requirements for AI and we needed a plan that involved all our staff"
Meanwhile, Mark says JE Dunn took a different path. Their AI rollout started with a top-down leadership buy-in and a structured internal committee to manage adoption at scale.
"We went full in with Copilot and bought licenses for most of the people in our company,” he explained. “We had an executive committee, so all of our leadership got bought in on Copilot, and we created an internal committee that was not run by IT."
This committee led a highly structured ideation process, holding biweekly meetings to collect and prioritize use cases from operators. This process allowed JE Dunn to thoughtfully curate and validate ideas before passing them over to IT for implementation.
As for their most-used AI tools? The answers span both commercial platforms and custom-built solutions.
Over at Hazen, Jamie said they use Copilot for speeding up day-to-day tasks like meeting minutes and QA/QC processes, while ensuring work quality and IP protection remain top priorities. He emphasized that even when companies attempt to restrict access to generative AI, employees are likely finding ways to use it. So, it's better to lean into inclusive and responsible adoption strategies.
Mark is a big fan of artificial intelligence, saying that JE Dunn has fully integrated it into its workflow. They blend internal ERP and Autodesk data into a centralized data lake using Microsoft Fabric, then layer machine learning and Copilot to power their AI-driven insights.
"We are using AI, but we are blending all of our internal ERP data and our Autodesk data—all of that is put into fabric for our data lake. And then we're using internal data models for our machine learning and Copilot for AI."
Then there's Construction IQ powered by Autodesk AI. Brandon with Rosendin said the firm has seen a lot of benefits from the tool, including early risk detection across quality and safety workflows.
"You go to the Insight side of Autodesk Construction Cloud, and you can see how the AI is pulling all the quality data and safety data and then assigning risk to it based on what information was entered. I think that's definitely a step in the right direction to be more proactive and see those early warnings so we can take action on them."
Being such a versatile technology, AI has a wide range of use cases that touch nearly every phase of a construction project. Here are some of the ways it's being applied today.
Some firms are leveraging AI to accelerate onboarding and reduce the knowledge gap for new team members.
As Mitch put it, "We're thinking about it in terms of the onboarding cycle for new project team members or new projects. When you bring somebody into a project, how quickly can you get them up to speed? When leveraged well, AI can shorten the onboarding cycle."
Mark echoed this and added, "Using AI and bots to provide the onboarding information to new employees is something that we're heavily looking at. Right now, we're doing a few things. We're taking large manuals and a lot of the PDFs and content from our intranet and essentially providing it to a Copilot bot to allow the creation of a virtual agent that new employees can interact with."
"All that information is already out there, but it's hard to onboard these new employees, especially with the pace at which we're growing."
Brandon brought up the value of using AI to enhance jobsite safety and quality assurance through photo and video recognition tools.
"Quality and safety for me go hand in hand. So, we need to be able to utilize AI to recognize video or pictures as far as any potential safety risks that we can mitigate early. There's also code compliance; having the AI understand where any code violations would help out, especially with the ever-growing workforce."
AI also has several use cases in the efficiency department—particularly when eliminating repetitive, manual tasks that consume valuable time. In fact, according to a McKinsey study AI could boost construction productivity by up to 20% via better project planning and management of resources.
AI can help with "the repetitive, the tedious, the non-value added tasks that we have to do all day," said Mitch.
"Think AP entry, insurance certificate review, meeting minute optimization, or pre-task planning. These things can be massively disrupted with AI tools."
Deciding which AI solution to implement is critical—but equally important is making sure the underlying data is clean, accessible, and properly structured to enable AI to deliver real value.
At Fortis Construction, Mitch explains how they’re navigating this balance. “Tactically for us, this means we're working a lot on the balance between structured and clean data versus unstructured data.” he says. “Historically, every AI or machine learning process required a pretty healthy amount of structured data, but generative AI tends to prefer unstructured data to structured data. And so it starts to change how we think about it."
This shift in thinking underscores the importance of not just collecting data but managing it thoughtfully.
At Rosendin Electric, Brandon highlighted that standardizing data collection and building a centralized "data lake" has become a core part of their AI readiness strategy.
"For us, it's really standardizing our processes to understand what data we want to capture, and then building those databases and pulling all that data in,” he explained. “So we have one giant data lake from which we can pull, and the AI can grab that and use that data down the road."
Ultimately, without strong data foundations, even the most advanced AI tools will struggle to produce meaningful results.
While the benefits of AI in construction are clear, industry leaders agree: adoption can't outpace caution. Ethical considerations—especially around data ownership and accuracy—should be considered as firms scale their AI efforts.
When it comes to AI, one of the biggest questions is: who owns the data, particularly when client-owned 3D models are uploaded into shared environments?
Jamie acknowledged this complexity, especially in engineering. "Historically, the electronic artifacts create our product, not the owners."
Rather than fighting to retain ownership, Jamie said their firm focuses on using data collaboratively to strengthen client relationships and drive progress.
"We're less concerned about owning it and more about making sure it's good data that can push us forward together," Jamie added.
Meanwhile, Mark chimed in about the operational side of this challenge. To comply with CMMC (Cybersecurity Maturity Model Certification) requirements, "We have to create separate enclaves for our Autodesk data, any attachments, and keep those completely isolated... Not everybody can see that data set."
Looking ahead, the panel agrees that Autodesk Bridge’s functionality—which allows teams to share specific data across separate projects—could help strike a better balance between ownership, security, and collaboration.
It's no secret that AI platforms can hallucinate—i.e., generate responses that sound accurate but are incorrect or fabricated.
To this, Mitch explained that his team actively tests every AI tool they encounter to evaluate not just functionality, but how often—and how severely—they hallucinate.
The firm also has some guardrails, including the policy that Copilot is the only AI tool officially supported.
And, of course, they always have a human review component in all AI-assisted workflows. Doing so ensures that AI outputs are accurate, trustworthy, and safe to use in high-stakes environments like construction.
As Mitch pointed out, "You have to validate anything that is generated by these tools. That human in the loop is so critical. Full stop."
For Brandon, it all goes back to mutual trust. “Can we trust AI yet? Is it ready for us to trust? Especially when you're building in the field and we've been doing it this way for a long time, how does AI know better than what we know because we’ve experienced it?”
While we may not be able to trust AI’s output in blind faith, it’s more important to recognize and work with its limitations. “It's just a matter of understanding where it's at, learn how to leverage it, and build that trust for those teams to be able to get those efficiencies. It's not there to replace them, it's there to enhance them and make them better,” Brandon added.
AI is already reshaping construction workflows—and the firms leaning in with intention, structure, and a healthy dose of curiosity are seeing the biggest gains. But success requires more than just adopting new tools; it demands clear data strategies, ethical safeguards, and a people-first mindset.
Want to see how Autodesk is helping teams use AI more intelligently and responsibly? Explore Autodesk AI to see how you can leverage it in your workflows.
