19.1% of US workers AI replaced have been asked back
The short answer
Almost one in five American workers who say AI replaced their job have since been asked to come back, according to a survey run by workforce assessment firm SHL in June and July 2026. The figure is 19.1 per cent in the United States and 16 per cent when British respondents are included. The workers who returned are not coming back quietly: 51 per cent demanded a substantial pay rise, 42 per cent wanted written job-security guarantees, and 40 per cent wanted the employer to say plainly where AI is used.
What the survey actually measured
SHL commissioned Propeller Insights to survey more than 1,000 workers in the US and more than 1,000 in the UK between June and July 2026. The question was narrow: have you personally been replaced by AI, and were you afterwards asked to return to the job. In the US, 19.1 per cent of respondents answered yes to both. Combine the two countries and the number falls to 16 per cent, which suggests the effect is stronger in the US labour market than the British one.
Fortune reported the findings on 9 November 2026. The story is not that AI has stopped working. It is that a slice of the cuts made in the name of AI were premature, and undoing them costs more than the cut saved.
The numbers worth remembering
The 19.1 per cent is the headline. Three other figures explain why this is a labour story rather than a technology story.
- 24 per cent of workers aged 18 to 24 said they were replaced and then asked back. The youngest cohort has the least protection and the most exposure to entry-level automation, and it is also the group being recalled fastest.
- 32 per cent of US hiring managers whose companies cut roles for AI told a separate Robert Half poll that the same or similar positions were later reinstated. Managers and workers are describing the same reversal from opposite sides of the table.
- 51 per cent of returning workers wanted a significant pay increase before agreeing to come back, 42 per cent wanted job-security guarantees in writing, and 40 per cent wanted transparency about AI use. Only a minority came back on the old terms.
Why the reversal happens
Three patterns show up in the survey data and in the way companies talk about AI cuts. None of them is about the model being bad. All of them are about the work around the model.
First, the workflow was never fully costed. A tool can draft a support reply or a first-pass design, but somebody still has to check it, handle the exceptions, and take the blame when it is wrong. Companies that removed the checker found the error rate went up in ways customers noticed. Rehiring the person who used to catch those errors is cheaper than losing the account.
Second, hiring back is slower than letting go. A laid-off worker knows the systems, the customers and the unwritten rules. A replacement needs months. When a company realises in week six that the work is not getting done, the fast fix is a phone call to the person who left, not a job advert.
Third, AI cuts became a signal. Once a company announces layoffs "because of AI", it has told the market that its headcount was previously inflated. Reversing partly is a way to keep the productivity story while admitting the arithmetic was wrong.
What the returning workers are asking for
The demand list is the interesting part of the survey, because it is a list of what workers learned. A pay rise is the obvious ask: if the company is calling, the worker has leverage. Written job-security guarantees are less obvious and more revealing. Workers are asking to be protected from being cut a second time for the same reason. Transparency about AI use closes the loop: nobody wants to train the system that is meant to replace them without knowing it.
These are not demands from a position of strength in a normal market. They are the terms of people who have already been through one round and will not do it twice for the same money.
The catch in the data
This is a self-reported survey of just over 2,000 people across two countries, run by a company that sells workforce assessment tools. It measures whether people were asked back, not whether they accepted, and it does not say what share of AI-related layoffs were later reversed across the whole economy. The Robert Half poll covers hiring managers, not workers, and it asks about "same or similar" roles, which is a wide net.
Treat the direction as real and the size as approximate. If a fifth of people who say AI took their job have been asked to return, the "AI replaced us and never looked back" story is too simple.
What this means if you are the one being replaced
Three practical points follow from the survey. Keep a record of what the job involved, because the recall conversation will be about specific work, not titles. Ask what changed, in writing, if the company says AI will do the job. And if a former employer calls, remember that the recall rate on the other side of the table is high enough that you can negotiate terms instead of accepting the old contract.
For employers the lesson is narrower and more expensive. An AI cut that is later reversed costs the redundancy payment, the rehiring cost, the lost knowledge, and now a pay rise on top. The survey does not say AI does not work. It says the accounting usually leaves out the part where people check the output.
Where this goes next
The recall pattern changes what a good AI business case looks like. If a fifth of the people you replaced are needed again within months, the honest comparison is not salary against software subscription. It is salary against software plus error rate plus the cost of rehiring plus the premium the returning worker now charges.
It also changes the legal picture. A company that calls people back after an AI redundancy is doing something useful and slightly awkward: it is admitting, in payroll records, that the automation did not remove the role. That admission matters when the next case reaches a tribunal, and it matters more in places where AI-driven dismissal is starting to have its own rules. The Philippines bill filed this week and California's WARN changes are the first versions of that idea.
Questions people ask
How many US workers have been asked back after AI replaced them?
19.1 per cent of US respondents in the SHL survey said they had been replaced by AI and were later asked to return. Including UK respondents, the figure is 16 per cent.
Who ran the survey?
SHL, a workforce assessment company, commissioned Propeller Insights to survey more than 1,000 workers each in the US and UK between June and July 2026. Fortune reported the results on 9 November 2026.
What do returning workers want?
51 per cent asked for a significant pay increase, 42 per cent wanted written job-security guarantees, and 40 per cent wanted transparency about how the employer uses AI.
Does this mean AI layoffs are being reversed across the economy?
No. The survey measures whether people were asked back, not the share of all AI-related layoffs that were reversed. A separate Robert Half poll found 32 per cent of US hiring managers reinstated the same or similar roles, which points in the same direction but is not an economy-wide count.
Which age group is most affected?
Workers aged 18 to 24: 24 per cent said they were replaced by AI and later asked to return, higher than any other group in the survey.