We’ll start with a confession.
A few years back, at the forefront of the AI frenzy, one of our team members asked a chatbot for help with a research project for our annual Women in Construction Month celebration. The request was simple. “Find women who made meaningful contributions to the construction industry, and give their stories, accomplishments, and dates of those contributions.”
The report came back polished and inspiring, with four women, four careers, and four sets of accomplishments. Polished. Inspiring. And wrong. Two of those women never existed. The chatbot had invented their names, their stories, and their milestones.
Fortunately, our Belco teammate caught the problem before anything went out. That catch has become a hard-won lesson. Initially, she reports feeling genuinely shocked and couldn’t shake the feeling that AI was an untrustworthy tool. Particularly in research or data deployment. But that was back when ChatGPT was churning out mostly distorted images with badly spelled words.
AI has improved hand over fist since then, and so has our understanding of how to ask it for better work. It can give real time back to people buried in admin work. However, that payoff only comes to teams willing to climb the learning curve and keep real, knowledgeable industry experts in charge, while remaining authentic to one’s brand and company values.
The Fear is Real. So is the Gap.
For so many people, AI still feels less like a tool and more like a threat. In a Pew Research Center survey published February 25, 2025, 52% of U.S. workers said they feel worried about AI’s future impact on the workplace. Only 36% said they feel hopeful. Gallup polling released September 15, 2026, found 27% of workers worry technology could soon make their jobs obsolete. That’s a new high, up from 20% last year.
Those worries deserve an attuned listening ear. Still, the early evidence points to a more measured integration of change. In a Gallup study published June 17, 2026, just 1% of currently laid-off workers named AI or automation as the primary cause. Gallup notes AI may still play an indirect role. Meanwhile, Brookings reported in March 2026 that 73.8% of built-environment workers hold jobs where AI is more likely to complement them than replace them.
Leadership matters here, too. A Gallup analysis from September 9, 2026 found workers who strongly agree they’re treated with respect, worry less about losing their job to AI. Our take on this? How a company introduces AI can massively shape how its people experience it.
Where AI Is Actually Showing Up
Across our industry, AI is landing in the back office first. The AGC and Sage 2026 Construction Hiring and Business Outlook came out January 8, 2026. It found that 45% of contractor firms use AI for office and administrative work. Fewer use it for estimating (23%) or design and preconstruction (20%). Among home builders, NAHB reported in August 2025 that about 20% use AI to create advertising and marketing materials. NAHB added that most builders are “still on the sidelines.”
Some of the most practical examples sit right in our corner of the supply chain:
- Takeoffs from plan sets. In June 2026, one of the nation’s largest LBM distributors became the first partner for a new AI estimating tool. It turns builder plan sets into 3D models and material takeoffs.
- Order entry. A sixth-generation Southeast dealer tested AI that reads emailed POs, spreadsheets, and jobsite photos, then drafts the order. An employee approves every draft before it reaches the dealer’s system.
- Builder operations. On its April 2026 earnings call, Taylor Morrison reported “more than a dozen AI-powered applications in production across finance, sales, purchasing and customer experience.”
Even so, most of the industry is still early. In the RICS AI in Commercial Property and Construction Report, published August 18, 2026, about two-thirds of construction respondents use AI somewhere in their work. Yet only 19% use it regularly in specific processes. Fewer than 1% have fully integrated it across their organization. In other words, nearly everyone is still learning.
The Tools Doing the Heavy Lifting
So which tools are people actually using? For now, general-purpose assistants lead. In RSM’s survey published July 29, 2025, construction respondents using generative AI named ChatGPT (83%), Microsoft Copilot (51%), and Google Gemini (43%). A July 2026 Mastt survey of construction project managers showed the same top two, with Anthropic’s Claude in third. AI built into construction software platforms barely registered.
Both samples are small, so read this as a direction rather than a ranking. Still, the pattern is useful. Most teams start with an everyday assistant, not a major software purchase.
The Learning Curve Is the Real Work
Our team member’s experience wasn’t unusual. A global study by KPMG and the University of Melbourne, published April 29, 2025, found 66% of employees rely on AI output without evaluating its accuracy. More than half (56%) said they have made mistakes in their work because of AI. Only 47% had received any AI training.
Seasoned professionals aren’t immune, either. In October 2025, Deloitte agreed to partially refund the Australian government for a A$440,000 report. It contained a fabricated quote from a federal court judgment and references to non-existent research. That’s the kind of cautionary tale that can either stall AI adoption or sharpen demand for ongoing training and for keeping not just a “human in the loop,” but an expert in charge.
Also, and this is an important piece of the puzzle, mistakes eat into the time AI saves. Workday research released January 14, 2026, found rework consumes nearly 40% of AI time savings.
The answer isn’t avoiding AI. It’s building better habits:
- Ask for sources, then check them. A confident answer isn’t a verified one.
- Give rich context and clear guardrails. Better prompts produce better first drafts.
- Stress-test the output. Decks, data analysis, and research summaries all deserve a second look.
- Keep a person in the approval seat. That dealer’s order-entry setup we noted above gets this exactly right.
How We're Approaching It at Belco
At Belco, we started by listening. We asked our team where AI could take weight off their workload, and we’re building training around their answers.
For now, our focus stays inside our own walls. We use AI to speed up research, analyze market trends, sharpen marketing and SEO research, and draft first versions of presentations. Every output gets stress-tested before anyone relies on it.
Our goal isn’t to replace people. It’s to clear the bottlenecks that slow good people down. When routine work moves faster, our team gains headroom to think ahead, solve problems together, and serve customers with more focus.
Questions You Can Take to Your Team
As part of our rollout, we asked our team a set of questions. Here are a few worth asking anywhere AI is on the table in your organization:
- Which tasks take the most time in your week without needing your expertise?
- Where does work stall while it waits on someone or something?
- Which AI tools, if any, are you already using, and for what?
- What would you want AI to take off your plate? What would you never hand over?
- What worries you most about using AI in your role?
- What kind of training would help you feel confident: hands-on practice, written guides, or one-on-one help?
- How confident are you that you would know when an AI tool you’re using has made a mistake?
- What training method or style would be most impactful: one-on-one guidance, a self-paced course, or group learning with some hands-on assistance?
Their answers are shaping where we start and how we train.
The Work Still Belongs to People
Our industry doesn’t have hours to waste. In January 2026, ABC estimated that construction needs to attract about 349,000 net new workers this year. In the AGC and Sage outlook, 82% of firms reported trouble filling hourly craft positions.
AI won’t frame a wall, mill a board, or earn a customer’s trust. What it can do is hand some highly burdened, and highly skilled people back some of their time. That time is worth the learning curve.
