Firms That Replaced Human Workers With AI Are Now Having to Rehire Staff Quickly

Companies that reduced their workforces after adopting artificial intelligence are beginning to recruit people for roles they had expected technology to handle, as employers discover that chatbots and automated systems cannot always replace human judgement, experience and oversight.

The trend, known as the 'AI boomerang', is illustrated by Swedish payments company Klarna and US carmaker Ford. Klarna resumed recruiting human customer-service representatives after relying heavily on an AI chatbot, while Ford hired or promoted experienced engineers after AI-based quality-control tools fell short of expectations.

Research by staffing firm Robert Half, reviewed by Fast Company, found that 32 per cent of 2,000 surveyed US hiring managers said their organisations had eliminated a role or let an employee go primarily because of productivity gains from AI or automation, only to rehire for that exact role later.

Klarna Reconsiders AI-First Customer Service

Klarna, the Swedish buy-now-pay-later provider, began recruiting human customer-service representatives again in 2025 after relying heavily on an AI chatbot to handle customer enquiries.

Chief executive Sebastian Siemiatkowski acknowledged that the company's reliance on the chatbot to reduce costs had gone too far. Klarna had previously promoted the system's ability to handle customer-service work at a scale comparable with hundreds of human agents.

Klarna has not abandoned AI, continuing to use automated systems for routine enquiries while retaining human representatives for more complicated problems.

Customer complaints can involve unusual circumstances, emotional responses or decisions requiring judgement. Such interactions may be difficult to resolve through automated systems alone.

Ford Retains Experienced Engineering Skills

Ford provides a different example of the limits of automation. In June 2026, the automaker said it had hired or promoted 350 veteran engineers over the previous three years after AI tools used for quality control did not perform at the expected level.

The engineers included former Ford employees and people who had worked for suppliers. They helped identify defects, investigate problems and strengthen quality control.

Ford has not said that all 350 engineers were previously dismissed because of AI. It is not a straightforward case of mass redundancies followed by rehiring, but shows the continuing need for experienced employees to supervise and improve automated systems.

A Ford spokesperson said AI had contributed to quality gains when combined with the technical expertise of its engineering teams.

Employers Discover the Limits of Automation

The examples point to a wider reassessment, although the scale of the trend remains uncertain. Robert Half found that employers most often cited institutional knowledge and context, greater human oversight and quality control, and lower-than-expected productivity gains when explaining why roles had been restored.

Among hiring managers reporting that their organisations had rehired, 40 per cent cited institutional knowledge or context that AI could not replace, 38 per cent pointed to greater-than-expected human oversight or quality-control needs, and 35 per cent said productivity gains were smaller than anticipated.

The survey does not mean that one-third of all companies using AI have reversed their job cuts. It reflects responses from 2,000 US hiring managers and should not be treated as a measure of every employer's experience.

Experts say businesses can overlook the informal work employees perform beyond their formal job descriptions. Staff may resolve unexpected problems, maintain customer relationships, share knowledge and build trust within teams. Those contributions can be difficult to identify when companies assess roles mainly by the tasks AI can automate.

Brooks Holtom, a professor of management at Georgetown University's McDonough School of Business, has argued that organisations often fail to understand their workflows fully before transferring responsibilities to AI systems. That can leave gaps in processes previously managed through employees' experience, judgement and collaboration.

AI Is Not Necessarily Replacing Entire Jobs

The cases do not establish that AI is failing across the economy or that companies are abandoning automation. Rather, they illustrate the difference between automating individual tasks and replacing an entire job.

Microsoft, Barclays Bank and supermarket chain Kroger were among companies announcing AI-related workforce reductions in 2026. Those announcements do not prove that every job cut was directly caused by automation.

For businesses, the reversals carry a practical warning: reducing headcount before testing AI in real working conditions can create additional costs if roles later need to be restored. Rehiring may also be difficult if former employees have moved on or are reluctant to return without better pay, clearer responsibilities or greater job security.

The emerging lesson is not that human workers and AI must compete. In some workplaces, automation may be most effective when it handles repetitive tasks while people provide judgement, accountability, technical knowledge and customer care.