Nearly 1 in 5 US Workers Say AI Replaced Them, Then Asked Them Back

What happened

Talent assessment firm SHL has published survey results describing what it calls "AI rebounding", a sequence in which an employer replaces staff with artificial intelligence and then asks the same people to return. Fortune reported the findings on 9 November 2026. The survey was carried out by Propeller Insights for SHL, and the sample included 1,029 employed US respondents. According to a report cited by Cornford and Cross, SHL published the results on 29 October 2026.

In the United States, 19.1% of respondents said they had been replaced by AI and later asked back by the same employer. Across the US and UK combined, the figure is 16%. Among respondents aged 18 to 24, the share in the US rises to 24%, according to Fortune's report of the results.

The figures are self-reported. The survey asked people how they understood their own experience, and the phrase "replaced by AI" leaves room for different readings. The five-point margin of error means the estimate for the US subgroup is less precise than the full sample, as Cornford and Cross note. The survey does not show how many jobs AI took over, and it does not show whether employers rehired at higher cost.

What returning workers asked for

SHL asked people who had been asked back what they would need before agreeing to return. The answers are set out below.

According to HR Dive's coverage of the same report, the three conditions are the terms returning workers said they would require. The survey measures stated expectations. It does not confirm that any employer met them.

Why it matters for businesses that automate work

SHL's global senior vice president of product and design, Lucy Beaumont, told Dataconomy that companies were making fast decisions about workers based on an incomplete view of what those employees could do with and without AI. Her wording describes the pattern as a company automating a role before it has assessed the human capability inside that role.

Gartner predicts that by 2029, 30% of workers terminated or replaced by AI will be rehired, often at a higher cost, as HR Dive reported. That is a forecast, not a measured outcome, and it sits alongside a growing count of AI-linked cuts. Fortune reports that outplacement firm Challenger, Gray & Christmas counted 121,401 announced job cuts where employers cited AI through October 2026, more than double the 54,836 announced layoff plans for all of 2025.

The practical point is cost and knowledge. A role that is cut and then refilled can cost more than the original, because the returning person now has leverage. The knowledge the role carried may also be gone before anyone notices it was needed.

What changes for a business that automates work, in practice

The survey does not argue against automation. It shows that the decision to remove a role is where errors become expensive. For a five-person office or a group with thousands of staff, the steps are the same:

1. List what each role does beyond its written tasks. Include exceptions, customer history, sign-off authority and the informal checks that catch mistakes.
2. Run the new tool alongside the person for a period you set in advance. Log every case where the person corrected or overrode the output.
3. Keep a human review step for exceptions, refunds, disputes and anything with a legal or financial consequence. Automation handles repeated steps well when the inputs are consistent. It does not replace judgement on unusual cases.
4. Tell affected staff, in writing, which tasks will be automated and which will stay with them, before the change. The 40% figure in the survey suggests this is a point people raise.
5. Do not cut a role on the strength of a tool that has not been tested on your own data. If a role is to be reduced, the notice, consultation and severance rules of the country where the people are employed apply. Agree the change with the employee or check it with a lawyer first.

None of this requires a large project. It requires a written list of tasks, a test period and a record of what the tool got wrong.

What to watch next

Three things will show whether the pattern grows. The first is whether SHL repeats the survey with the same question wording, which would allow comparison over time. The second is whether Challenger's monthly counts of AI-linked cuts keep rising. The third is whether Gartner's 2029 forecast of rehiring begins to appear in published employer data. Until then, the 19.1% figure should be read as one survey's record of what respondents experienced, not as a measure of how often employers have reversed their decisions.