Half the jobs cut for AI are coming back. Under new titles.
Through June of this year, AI was named as the reason for 101,743 announced job cuts in the United States. That’s according to Challenger, Gray & Christmas, who have tracked layoff reasons for decades. It’s nearly double the 54,836 they logged for all of 2025.
Here’s the part that didn’t make the headlines.
Of the executives who made those cuts, 55% now say it was the wrong decision.
Not “we’re monitoring the situation,” or “results have been mixed.” Wrong decision. Forrester surveyed more than 1,100 business leaders and that’s the number that came back.
Gartner went further, forecasting that by 2027, half of the companies that cut customer service headcount for AI will rehire people to do similar work, under different job titles.
New titles for everyone!
It’s already happening
A February survey of 600 HR professionals who ran layoffs in the previous twelve months found that more than a third of their companies had already rehired over half the roles they eliminated, most of them inside six months.
Six months. That’s not a market correction, it’s barely long enough to finish the severance paperwork.
And nearly a third of those HR leaders reported losing critical skills and expertise when those people walked out. Which is the sentence that should stop you cold, because it means the org chart came back and the knowledge didn’t.
Ford spent 900 cameras worth of confidence to learn this
My favorite version of this story is Ford.
They deployed roughly 900 AI cameras for quality control on the line. Machine vision, running defect detection, doing the job that experienced technicians had been doing with their eyes and about twenty years of knowing what a bad weld sounds like.
You can see where this goes… the cameras started missing things that the technicians hadn’t.
Ford rehired somewhere between 300 and 350 veteran engineers. Their VP of Vehicle Hardware Engineering said: the technology is only as good as the data used to train it, and Ford misjudged what AI could do alone.
Then Ford went and became the top mainstream brand in the JD Power 2026 Initial Quality Study.
So this isn’t a story about AI failing. Ford still runs the cameras. It’s a story about what the cameras couldn’t see, and about who could see it, and about the eighteen months Ford spent finding that out the expensive way for all of us to learn from!
What actually got cut
I’ve spent a lot of hours in rooms where a process gets documented for the first time. Every single time, the useful information is not in the SOP that already exists. It’s in the person who says “yeah, that’s what the document says, but if you do it that way in July the humidity screws it up, so what we actually do is…”
That’s the reason behind everything we do. Nuance.
And it’s invisible on a headcount spreadsheet. It has no line item. Nobody’s resume contains Knows Why We Stopped Doing It That Way In 2019. So when a company builds a business case around AI handling a function, the function it’s modeling is the documented one. The one in the flowchart made in 2006.
Then the people leave, and the company discovers how much of the operation was running on the undocumented version.
My dad has been an engineer for most of my life. He’s close to retiring. There is a genuinely alarming amount of load-bearing information in that man’s head that exists nowhere else, and every company I work with has three or four of him.
Nobody’s replacing that with a chatbot, but plenty of people are laying it off on a promise that I don’t see AI being able to fulfill.
The other number
Before this reads as an anti-AI piece, one more stat.
Forrester also found that more executives expect AI to increase their headcount over the next year (57%) than expect it to decrease (15%).
The productivity gains are real. Stanford’s synthesis of the research puts software development around 26%, customer support at 14 to 15%.
The problem was never whether AI works.
Why the math went wrong
The leap from “AI can assist this” to “AI can replace this,” made in a quarter, on a slide, by someone who had never done the job.
In my experience in build ROI cases for AI initiatives, a lot of these choices were because doing the math by FTE headcount is faster/easier/cheaper than doing a time study and trying to calculate time saved. Which that time saved does carry a tremendous amount of hard/soft benefits/value.
There are dollars and cents tied to this time saving but you also get a really large soft benefit here. It looks like you are making the investment in the wellbeing of your team.
Sally from reporting now gets to say, “OMG I’m so glad I don’t have to do this one tedious part of the job anymore it’s saved me like 8 hours a week!”
She hated the tedious work. It’s more effectively and consistently done with an AI agent. Head count math can’t see Sally, just her salary.
What I’d actually do
If you’re staring at a function and wondering whether AI can absorb it, the question isn’t “what does this role produce.” It’s “what does this person know that isn’t written down anywhere.”
Find that out first. Capture it. Then automate.
Do it in that order and you get the efficiency. Do it in the other order and you get to pay a recruiter to find someone who knows what your last person knew, which by definition is nobody, plus 6 months of ramp up because that knowledge left the building with them.
Half of the AI layoffs are coming back. The knowledge is not. As I watch layoffs I really wonder who’s going to hurt the most here. Are some enterprises large enough to take the loss and bounce back, sure. Why not be forward thinking enough now to stop the loss in the first place? I hope that Directors and up are considering this.
So, here’s what I keep chewing on: what’s the thing your company does every day that only works because one specific person is still there? And do you have the foresight strong enough to capture it before they are gone?
Sources
AI-attributed job cuts (101,743 through June 2026; 54,836 in all of 2025; AI leading reason five consecutive months)
Challenger, Gray & Christmas, monthly job cut reports, 2026.
https://www.challengergray.com/blog/challenger-report-layoffs-fall-hiring-picks-up-ai-leads-for-fifth-straight-month/
Additional coverage: CNBC, June 5, 2026 — https://www.cnbc.com/2026/06/05/ai-is-now-the-leading-reason-companies-give-for-cutting-jobs-says-new-report-what-that-means-for-workers.html
Tech sector share, H1 2026: HR Dive — https://www.hrdive.com/news/tech-layoffs-surge-83percent-h1-2026-challenger-ai-disruption/824320/
55% of executives who made AI-driven cuts say it was the wrong decision; 57% expect AI to increase headcount vs. 15% who expect a decrease
Forrester, Predictions 2026: The Future of Work. Survey of 1,100+ business leaders.
Reported via Capitol Technology University — https://www.captechu.edu/blog/ai-and-job-replacement-new-study-finds-surprising-correlations
Secondary source. Link the Forrester report directly if you have access.
Half of companies that cut customer service headcount for AI will rehire for similar work under different job titles by 2027
Gartner, 2026 forecast.
Cited across coverage of the Forrester and Challenger data. Link Gartner’s own release if you can locate it.
More than a third of companies rehired over half the roles they eliminated, most within six months; nearly a third lost critical skills and expertise
Careerminds, February 2026 survey of 600 HR professionals.
Reported via Curiouser.AI — https://medium.com/@curiouser.ai/the-great-ai-layoff-boomerang-68e38c88fa7d
Secondary source, and the weakest sourcing in this piece. Verify against the Careerminds original before publishing.
Ford: approximately 900 AI quality-control cameras, 300 to 350 veteran engineers rehired, comments from Charles Poon, VP of Vehicle Hardware Engineering
Forkast — https://forkast.news/the-ai-rehire-correction-paradox-87k-cuts-ytd-but-a-third-of-companies-are-already-reversing-course/
Ford ranked top mainstream brand, J.D. Power 2026 U.S. Initial Quality Study
J.D. Power, 2026 Initial Quality Study.
Productivity gains: approximately 26% in software development, 14 to 15% in customer support
Stanford Institute for Human-Centered AI, AI Index Report 2026, economy chapter.
Cost of a new hire (if you keep the SHRM figure)
Society for Human Resource Management, cited via ASCM — https://www.ascm.org/ascm-insights/beyond-hiring-why-upskilling-and-reskilling-is-the-key-to-talent-retention/
