Companies that laid off employees to implement AI are increasingly calling those same people back when automation fails to meet expectations: talent analytics provider SHL described this reversal as the AI rebounding trend in its study “Talent Mobility at Work in 2026: The Hidden Workforce.” The survey showed that a significant portion of workers in the US and UK have already witnessed such “boomerangs” in their companies. Laid-off workers are willing to return, but on their own terms: for higher salaries, written job security guarantees, and transparency regarding AI use in their roles.
What happened
On October 29, 2026, SHL published the study “Talent Mobility at Work in 2026: The Hidden Workforce,” introducing the term AI rebounding: employers fire people to automate their tasks with AI, then rehire them months later when the technology fails to deliver expected results. The findings are based on a Propeller Insights survey conducted in June–July 2026 among 2,036 working adults in the US (1,029 respondents) and the UK (1,007 respondents). 19% of American and 13% of British participants reported personally encountering the return of employees laid off for AI; among 18–24-year-old workers in the US, this figure reaches 24%. Return conditions were strict: 51% of such workers demand a significant salary increase, 42% demand written job security guarantees, 40% demand full transparency on AI use in their roles, and 8% refuse to return under any circumstances.
Context
The “boomerang employee” phenomenon—returning to a former employer after being laid off—has long been known in HR practice, but CNBC notes that AI-driven layoffs have spawned a new wave: previously, people returned more often by their own initiative, but now employers are inviting them back because automation has not paid off. The hype around AI agents is accelerating these decisions: executives are promised quick savings on personnel, while companies are not in a hurry to assess employees' actual skills and the real automatability of tasks before laying them off, according to SHL. Against this backdrop, the organization is betting on internal mobility: 47% of workers would prefer retraining and a role change within the company over being laid off. When a person is cut for AI, trust in the employer collapses, and SHL calls this the destruction of the “psychological contract”—which is why any return turns into a negotiation over new terms.
Why this matters for the industry
For the industry, the study shifts the conversation on AI automation to the full cost of error: a failed implementation costs not just licenses, but also the increased price of re-hiring the same people, who now negotiate their return. A direct implication for production teams is to start any “replace the role with AI” initiative with a measurable baseline: which tasks the model actually covers, what share of work remains human, and how a rollback is structured while retaining expertise during the transition. The survey also validates demand for a separate layer of tools that would pre-calculate the real automatability of tasks and the cost of error, so HR platform and eval analytics vendors will likely try to convert this request into audits of AI's impact on the workforce. In a two-year horizon, if the trend is confirmed by independent data, the argument “we'll fire people, AI will do it all” will be devalued, and reliability-first automation with a hybrid model may become the norm, where an AI agent works as an assistant and layoffs occur only after the system's performance on real tasks is confirmed.
Why this matters for users
For readers working in places where AI is being actively implemented, these data provide a sober benchmark: one should not expect to return on previous terms; those laid off for automation are only brought back after a negotiation in which the employee now has a strong position. The study itself can be used as an argument in negotiations with an employer, primarily regarding transparency about which tasks in your role are being handed over to AI and what remains for humans. A reasonable step before any layoffs is to document skills that the model does not replace: knowledge of business context, judgment in non-standard situations, client work, and connections between processes. It is also worth tracking which tasks in your role are already being automated or planned for automation, to understand in advance where to deepen expertise and where to request retraining.
What is still unknown / limitations
The figures provided are self-reports from workers, not a measurement of AI system quality: 19% and 13% describe the share of people who encountered a “boomerang,” not the share of failed AI implementations, and the base frequency of layoffs “for AI” is not disclosed in the data, so the survey cannot be converted into a metric for automation failure. The study was commissioned by SHL—a talent analytics vendor interested in promoting corresponding services; the wording of questions, the sampling weighting scheme, and the exact definition of the situation “AI does not deliver expected results” are not detailed in open materials. The sample is limited to the US and UK, so transferring conclusions to other markets requires caution. Finally, the alignment of the trend with the technical limitations of current LLMs is an interpretation by analogy, not a conclusion of the study itself.
Sources
- SHL — press release “Workers Replaced by AI Would Demand Higher Pay and Greater Job Security to Return, SHL Research Finds”
- CNBC Make It — AI-driven layoffs are creating a new class of boomerang employees
- Hacker News — discussion of AI-driven layoffs are creating a new class of boomerang employees
Author
Look at AI, editorial team