Companies That Cut Jobs for AI Are Quietly Hiring Workers Back

Companies that announced job cuts over the past two years as a result of AI adoption are increasingly reversing those decisions quietly. A new survey from staffing firm Robert Half shows the problem is far from marginal: it affects nearly a third of American managers, who admit that automation could not handle the responsibilities it was assigned.

The scale of the problem

Robert Half surveyed nearly 2,000 American managers responsible for hiring. The result is unambiguous: 32 percent said they eliminated a position primarily because of AI adoption, then had to hire someone for the same or a very similar role. The breakdown by industry shows where automation failed most severely - in finance, as many as 44 percent of these decisions were reversed, in HR departments 35 percent, and in technology 32 percent.

A similar picture emerges from data compiled by analytics firm Orgvue, which found that 39 percent of business leaders admitted to laying off employees directly because of AI, and 55 percent of them later called that decision a mistake. That means that in many organizations, more than half of the cuts justified by artificial intelligence turned out to be decisions made too hastily, without a full understanding of what the model could actually do on its own.

Ford and its graybeards

The most striking example is Ford, which over three years reinstated or promoted 350 experienced engineers, internally nicknamed "graybeards," to fix quality problems that automated processes could not handle. The company acknowledges that AI tools generated specifications and analyses that looked correct but lacked the practical knowledge gained through years of work on the production line.

AI is fantastic, but it's only as good as the data it was trained on - Charles Poon, Vice President of Vehicle Hardware Engineering, Ford

The decision to bring back experienced engineers translated into a measurable result: in June 2026, Ford achieved its best score in the JD Power Initial Quality Study in 16 years. CEO Jim Farley said that restoring human oversight over key production stages brought hundreds of millions of dollars in warranty savings, with the company aiming for a billion dollars in annual savings.

A bank that backed down

Commonwealth Bank of Australia went even further and publicly admitted its mistake. In July 2025, the bank laid off 45 customer service employees, citing the deployment of an AI-based voice bot meant to take over a significant share of phone calls. The financial sector's trade union disputed the bank's figures on the decline in call volume, arguing they did not reflect reality.

On August 21, 2025, just a month after announcing the cuts, Commonwealth Bank reversed the layoffs. In an official statement, the bank admitted that not all business circumstances had been fully considered when making the decision. It is a rare case of a major financial institution openly acknowledging that a decision to cut jobs because of AI was made too quickly.

Why automation falls short

IBM's case illustrates a mechanism that repeats across many companies. The internal AI-based assistant AskHR resolves 94 percent of IBM employees' HR queries on its own, which is an impressive result in itself. The problem lies in the remaining 6 percent of cases, which require ethical judgment, familiarity with organizational context, or nonstandard decisions - tasks the system cannot take on without human involvement. IBM has announced it will triple hiring for entry-level positions in the US across all business units in 2026.

Robert Half analysts explain the pattern simply: after six to twelve months, it turns out AI handles about 60 percent of a given role's duties efficiently, but struggles with the remaining 40 percent, which requires experience, situational judgment, or knowledge built up over years. As a result, companies revert to a model of human-algorithm collaboration instead of full replacement.

What comes next for the job market

Research firm Gartner predicts that by 2027, half of organizations that justified layoffs with AI adoption will recreate similar positions, often under a different name, to avoid the appearance of a full retreat from their earlier strategy. This marks a significant correction from the narrative of two years ago, when many companies announced job cuts as proof of technological readiness and cost savings.

For Polish employers following the American and Australian experiences, there is a concrete takeaway: staffing decisions based solely on the claimed capabilities of AI models, without a pilot phase and measurement of actual work quality, carry a high risk of a costly reversal within a few quarters. The Ford and Commonwealth Bank cases show that the cost of fixing these mistakes can exceed the earlier savings from job cuts.