Technology workers who touch A.I. tools less than once a month are roughly three times as likely to have been laid off as colleagues who use them at least monthly, according to a Gallup study released this week. The predicted probability of layoff among infrequent users in the sector lands at 18 percent. Among regular users, it’s 6 percent.

The finding emerges from a February 2026 survey of more than 23,000 American workers, including 660 who said they were unemployed because their position had been eliminated. Gallup controlled for age, education, and sector, which means the gap isn’t an artifact of older workers in legacy roles. It’s a behavioral signal inside the industry that built the tools.

The pattern extends beyond tech. Across the full sample, non-users (defined as people who touch A.I. once a year or less) made up 62 percent of laid-off workers but only 50 percent of those still employed.

What’s stranger is what the displaced workers themselves said. Only about 1 percent attributed their dismissal to A.I. directly. Meanwhile, Challenger, Gray and Christmas found that roughly 40 percent of job-cut announcements last month named A.I. as the reason. Employers and ex-employees are telling two different stories about the same event.

“That surprised me the most. They didn’t just blame A.I.,” said Jim Harter, chief scientist for Gallup’s workplace management and wellbeing practices. Workers cited restructuring, cost-cutting, and economic conditions, the familiar vocabulary of every downturn since 2001.

The broader labor picture remains less dramatic than the tech-industry vibe suggests. In the first quarter, 45 percent of workers reported staffing at their employer was unchanged. The Bureau of Labor Statistics counted 5.5 million hires against 1.9 million layoffs and discharges in March 2026. Federal government workers reported the sharpest pressure, with nearly 38 percent saying their employer was letting people go, versus 17 percent in the for-profit sector.

Harter cautioned against the obvious managerial response, which is to start treating A.I. usage frequency as a performance proxy. “I don’t think that’s the right direction. The real bottom line is: are they more productive?”

That distinction will be ignored. The legibility of a usage dashboard, compared with the difficulty of measuring actual output, all but guarantees that “prompts per week” becomes a column in someone’s spreadsheet by Q3. The 2013-era wave of activity-tracking software taught the same lesson, and middle management learned nothing from it.

Sources