90% of executives say AI hasn't boosted productivity yet. That's the number to check before you cut an estimator's job
New research tracking thousands of firms found AI-linked layoffs get almost no reward from investors and often backfire on productivity, because the fear they create outweighs the efficiency gained. Here's what that means before a GC or sub cuts staff on the promise of an estimating or submittal tool.
Nine out of ten executives say AI hasn't moved the needle on their company's productivity yet — and the companies that laid off staff and pointed to AI as the reason got no reward for it. New research tracking more than 3,200 public companies found the average stock market reaction to an AI-linked layoff announcement was close to zero. Some of those cuts made things worse, not better, because of what the layoffs did to the people who stayed. If your firm is weighing whether an AI estimating, scheduling, or submittal tool means you can run leaner, that's the number to sit with before you act on it.
What did the research actually find?
Researchers built the picture from data most companies don't usually put side by side: millions of employee reviews, roughly 10,000 earnings-call transcripts, and hundreds of AI investment and layoff announcements spanning more than 3,200 U.S. public firms over about five years. Management teams talked about AI with consistent optimism on those earnings calls. That optimism had no meaningful statistical relationship to the company's actual productivity numbers. Separately, when companies announced AI-related layoffs, the stock market barely reacted — the average return sat close to zero, not the bump you'd expect if investors believed the cut was buying real efficiency.
Why would a layoff make an AI tool work worse?
Because the tool's payoff depends on the people still using it, and those people were watching. The research found that AI-tied layoffs measurably damaged the sentiment of remaining employees toward the technology itself — and that sentiment turned out to be one of the strongest predictors of whether a company got any productivity gain out of AI at all. Workers who'd just watched a colleague's role get cut and blamed on AI didn't lean into the tool. They avoided it, worked around it, or used it just enough to not get flagged. The layoff was supposed to be the efficient move. Instead it undercut the thing that was supposed to make the tool pay off.
What does this have to do with a mechanical sub or a mid-size GC?
Estimating and submittal-prep roles are exactly where AI tool vendors are pitching headcount savings right now — "cut your takeoff time in half," "your submittal coordinator can carry twice the volume." Those claims may hold up on your own work. But this research is a warning about the move that usually follows a good pilot: treating projected time savings as a green light to cut the role, before you've actually measured what the tool does on your jobs, and without thinking through what watching a colleague get cut does to everyone else's willingness to use the thing.
Before acting on a vendor's productivity claim, a few checks are worth running instead:
- Measure the tool on your own work, not the vendor's demo. Run it against real bids or real submittal packages for a full cycle and count actual hours saved and errors caught — not the vendor's benchmark numbers.
- Separate "the tool is faster" from "we need fewer people." A faster estimator can cover more bids or turn RFIs faster; that's often worth more to a sub's pipeline than the headcount saved.
- Watch what a cut does to the rest of the team. If a role disappears the week after an AI pilot launches, expect the survivors to treat every AI tool that follows with suspicion — which is the exact dynamic this research found kills the productivity gain.
- If you do reduce headcount, say what the freed capacity is for. A team that sees the tool absorb real work and gets redeployed to it responds differently than a team that just watched someone get let go.
We've flagged the underlying pattern before: enterprises are standing up new AI agents faster than most back offices can even assign an owner to them, let alone measure what those agents actually deliver. This research is the other half of that problem — companies making staffing calls on AI's promised output before they've measured its real one, then finding out the market and their own workforce didn't buy the trade either.
The tool might genuinely save your estimating team real hours. Prove it on your own numbers before you act on someone else's projection.
Before headcount comes off a bid or submittal team because of a new AI tool, run one full cycle of real numbers first — the market isn't rewarding the shortcut, and your own team is watching to see if you take it.
Construction AI Brief publishes three times a week. Subscribe at constructionaibrief.com.
- Has AI actually boosted productivity at most companies?
- Not yet, according to most of the executives who'd know. New research on corporate AI adoption found about 90% of executives report AI has not measurably boosted productivity at their company so far, even as many of those same companies keep investing in it and, in some cases, cutting staff.
- What happens to a company's stock price when it announces layoffs tied to AI?
- On average, close to nothing. Researchers tracking stock reactions to AI-linked layoff announcements across thousands of firms found the average market return was near zero — investors aren't rewarding the move as the cost-saving win executives often frame it as.
- Why would laying off staff after buying an AI tool hurt productivity instead of helping it?
- The research found AI-related layoffs damage the sentiment of the employees who remain, and that sentiment toward AI is one of the strongest predictors of whether a company actually gets productive use out of the tool. Workers who watched colleagues get cut over AI tend to resist or quietly avoid the tool themselves.
- What did the study measure to reach these conclusions?
- Researchers analyzed millions of employee reviews, roughly 10,000 corporate earnings-call transcripts, and hundreds of AI investment and layoff announcements across more than 3,200 U.S. public companies over about five years, comparing management's optimism about AI to measured productivity and market outcomes.
- Should a construction firm cut estimating or submittal staff after adopting an AI tool?
- Not on the vendor's ROI projection alone. Measure the tool's actual output on your own work for a full bid cycle or submittal season first, and if the workload drops, reallocate that person to work the tool created capacity for — like RFI turnaround or bid coverage — rather than eliminating the role outright.