Replaced by AI, then asked to come back
In the US, 19% of surveyed workers say they were replaced by AI and later asked back. That doesn't prove AI fails. It shows companies work in the wrong order.
Lees in het Nederlands →
By now everyone knows the headlines: a company cuts hundreds of jobs "because of AI". What you read far less often is what happens a few months later. That's when some of those people get a phone call asking whether they'd like to come back.
In an SHL survey of 2,036 workers in the United States and the United Kingdom, 16% say they were replaced by AI and later asked back by the same employer. In the US it is 19%, nearly one in five. SHL published the figures on 29 October 2026 and calls the phenomenon "AI rebounding".
Does that prove AI doesn't work? I don't think so. It does show that many companies do things in the wrong order. I'll walk through the numbers, the reasons, and how to do it the other way round.
What does the research say about being replaced by AI?
The survey was fielded by Propeller Insights in June and July 2026, commissioned by SHL, among 1,029 workers in the US and 1,007 in the UK. The key figures:
- 16% of all respondents say they were replaced by AI and later asked back by the same employer.
- 19% in the US, against 13% in the UK.
- 24% among 18 to 24 year olds. Young workers are hit most often.
Do read those numbers correctly. This is what workers report themselves, not data from employers. It is about people who were asked back: how many actually returned was not measured. And SHL sells talent assessment services itself, so it has an interest in the message that companies should know their people better.
On what terms do people come back?
Most are willing to return, but not on the old terms. According to SHL, this is what they ask for:
- 51% want a substantial pay rise.
- 42% want specific job-security guarantees.
- 40% want full transparency about how the company uses AI.
- 8% would not return under any circumstances.
That is what workers want. Whether they get it is another matter. According to The Next Web, research firm Forrester expects half of the layoffs attributed to AI to be quietly reversed, but often offshore or at lower pay. 55% of employers are said to regret their AI layoffs. The article doesn't say how Forrester arrived at those figures, so treat them as an estimate.
Why do companies bring people back?
Fortune cites a Robert Half survey from July 2026. In it, 32% of US hiring managers whose company cut positions because of AI say they later had to fill the same or similar roles again. The reasons they give:
- The role needed company knowledge AI could not replace (40%).
- Managing relationships (39%).
- Business demand increased (39%).
- More human oversight and quality control was needed (38%).
The best-known example is Klarna. In February 2024 the Swedish payments company said its AI assistant was doing the work of 700 customer service agents. In May 2025 CEO Sebastian Siemiatkowski admitted to Bloomberg that the company had gone too far, and Klarna started hiring people for customer service again. A spokesperson told Fortune that humans are needed for complex and sensitive cases.
Ford appears in the Fortune piece too: the carmaker hired and promoted 350 experienced engineers in June, after AI tools for quality control did not perform as expected.
How do you do it right?
Nearly all those reasons come down to the same thing: the company didn't know exactly what the person in that role was doing until they were gone. So turn the order around. Process first, people second.
- Start with tasks, not roles. Together with the people who do the work, write down which tasks their job consists of. AI replaces tasks. A role is always more than the sum of them.
- Automate one task and let human and AI run side by side for a while. That way you see where it goes wrong while someone is still there to catch it.
- Measure quality, not just cost. How many cases come back? How satisfied are customers? If you only look at the saving, you spot the problem when the customer has already left.
- See where the freed-up time is worth most. In the SHL survey, 47% say they would rather be retrained than lose their job if AI changed their role. So the willingness is there.
- Only then decide on headcount, and be honest about what you're doing. After a few months you know what AI really takes over. Then you decide on facts instead of on a promise from a sales demo.
My take?
These figures are now being used everywhere as proof that AI is overhyped. That's not how I read them. Klarna kept using AI in its customer service, alongside the people it brought back, and Ford itself says AI improved its quality. The technology does its job, just not the whole job.
The mistake is in the order. If you cut people first and only then discover everything they did, you pay twice: once for the layoff, and once more to bring the same people back at a higher price and with less trust. And the colleagues who watched it happen won't forget.
For a small or medium-sized business this is not some distant story. In companies like ours, the knowledge often sits with two or three people and is written down nowhere. My advice: go ahead and automate, the sooner the better, but do it together with the people who know the work. They know where AI will slip up long before you see it on a dashboard.
Frequently asked questions
What is AI rebounding?
It is SHL's name for what happens when a worker is replaced by AI and asked back by the same employer months later.
Does this mean AI doesn't replace jobs?
No. According to outplacement firm Challenger, Gray & Christmas, cited by Fortune, US employers announced 121,401 job cuts citing AI through October 2026, against 54,836 in all of 2025. Some of those are reversed, most are not.
Do these figures apply to Belgium and the Netherlands too?
We don't know. The SHL survey only ran in the US and the UK. I'm not aware of comparable figures for Belgium and the Netherlands.
Sources
- SHL: Workers Replaced by AI Would Demand Higher Pay and Greater Job Security to Return (press release, 29 October 2026)
- Fortune: 'AI rebounding': Nearly 20% of U.S. workers replaced by AI are being asked back (9 November 2026)
- The Next Web: Employers are quietly rehiring the people AI replaced (23 August 2026)
