Gartner just told a lot of executives something they don't want to hear. The research and advisory firm studied more than one million layoffs that happened in 2025 and found that jobs genuinely eliminated because AI made workers dramatically more productive made up less than 1 percent of all cuts. Yet AI got blamed for a lot more than that.
Gartner predicts that by 2029, nearly one in three employees laid off in the name of AI will need to be rehired, and it will cost companies more the second time. Tori Paulman, the Gartner analyst behind the research, put it plainly: AI is not taking over enough of anyone's job yet to justify letting large numbers of people go.
Gartner's team also found something worth sitting with. In the first half of 2025, 17 percent of layoffs attributed to AI were actually companies shifting money and staff away from struggling business lines toward new AI products, not layoffs caused by AI doing anyone's job. Paulman calls this AI washing: companies selling AI tools have an incentive to claim they use AI internally too, even when the real reason for the cuts is ordinary belt tightening.
Customer service is where this plan has already gone wrong in public. Klarna cut 700 support agents in 2024 and leaned on AI to cover the work, then started quietly hiring people back within about a year once service quality slipped. Ford, IBM, and Commonwealth Bank of Australia have all walked back similar cuts in engineering, HR, and support roles. The common thread is simple: customers do not like dealing with AI when something goes wrong, especially over the phone.
There is a second problem stacking on top of the first. The world's working age population is shrinking as birth rates fall and a large wave of experienced workers retires over the next several years. That means the pool of people available to rehire is getting smaller at the exact moment companies may need to reverse course. Fewer available workers plus rising demand for them equals higher wages and slower hiring, which is the opposite of what a company wants after already paying severance once.
None of this means AI is useless. Gartner's own separate tracking shows AI adoption is real and growing, and the firm expects AI to start adding jobs on net starting in 2028. The mistake is treating AI as a stand in for a person rather than a tool that changes how a job gets done. Gartner calls this the difference between AI as a replacement and AI as what it calls a toolmate, something that makes an employee better at their job rather than something that takes the job away entirely.
Other data backs this up. Accenture's Pulse of Change research found that even among companies seeing productivity gains from AI, a shrinking share can point to measurable business results, down to less than a quarter earlier in the year. A separate survey by Aptean found less than half of organizations consider AI essential to their core work at all.
For any business owner watching the layoff headlines and wondering if they should follow suit, the lesson from this research is straightforward. Cutting a role because a headline made it sound easy is a different decision than cutting a role because you tested AI on that specific work and it held up. The companies rehiring right now skipped that test. The ones avoiding a rehire in 2027 or 2028 are the ones doing it now.