Ford has acknowledged that it overestimated what AI could replace in product design and quality control, then spent the past three years rehiring 350 veteran engineers to close the gaps that followed.
According to Forbes, Charles Poon, Ford’s vice president of vehicle hardware engineering, said the company got the call wrong. “We mistakenly believed that if we brought in artificial intelligence and fed our design requirements into it, it would produce high-quality products,” Poon said.
The report says the problem was not that AI had no value. Ford misjudged which parts of the job AI could actually absorb.
This was not limited to one company. An international survey commissioned by organizational design software company Orgvue and carried out by Vitreous World polled 1,163 C-suite executives and senior decision-makers. It found that 39% had cut jobs after adopting AI, and 55% of that group later said the layoffs had been a mistake.
The same survey found that 23% of companies based those cuts on broad assumptions about AI capabilities instead of reviewing, role by role, what employees were actually doing each day.
Ford brought engineers back to fix what AI missed
Of the 350 people Ford rehired, some were former employees and others had already moved into supplier-side roles. Media reports referred to them as “gray beard” engineers.
Poon said the core issue was not the technology itself but the training data. Some of Ford’s most experienced engineers had already left before their knowledge was properly recorded. That meant the know-how they built over decades never made it into the design requirement documents, and never entered the training data either.
The report points to examples such as unusual sounds that only seasoned engineers could identify and assembly tolerances recognized through long experience. Those judgments were never formally written down, so AI had nothing to learn from.
Ford’s quality metrics improved after the rehiring effort. In the latest J.D. Power Initial Quality Survey, Ford ranked first among mainstream auto brands, its first return to the top in 16 years. Chief Executive Officer Jim Farley said lower warranty and recall costs had added up to “multiple hundreds of millions of dollars” in cost tailwinds.
IBM automated 94% of routine HR requests, but not the last 6%
IBM’s example came from human resources. Its internal AskHR system took over part of the company’s HR workload and handled about 94% of routine requests. The problem sat in the remaining 6%, where ethical judgment and exception handling were still needed.
IBM then said it would triple hiring for entry-level roles in the United States in 2026 across all business units. Speaking at a forum in New York, Chief Human Resources Officer Nickle LaMoreaux put it this way: “If we don’t continue investing in entry-level talent, what happens in three to five years?”
The report says IBM did not simply place people back into the same jobs. New HR hires step in when the chatbot cannot provide a sufficient answer, correct its output and communicate directly with managers. Junior software engineers are writing less routine code and spending more time in client conversations.
Commonwealth Bank of Australia reversed customer service cuts
Commonwealth Bank of Australia took a similar path. In July 2025, it cut 45 customer service roles, saying an AI voice bot had reduced weekly call volumes by 2,000.
The Financial Sector Union challenged that claim before a labor arbitration body, arguing that call volumes were actually rising. The bank, the union said, still needed customer service staff to work overtime and had team leaders answering phones themselves.
On Aug. 21 of the same year, the bank withdrew the layoffs, issued a public apology and paid back wages. In its statement, it said an initial assessment concluded the 45 roles were no longer needed, but it had failed to fully account for all relevant business factors. That error meant the roles were not, in fact, redundant.
Surveys point to costly reversals
Forbes columnist John Werner summarized the cycle as a pattern: a company announces that AI will replace a job category, cuts staff, waits six to 12 months, finds that AI can handle roughly 60% of the workload but not the remaining 40%, and then hires people back.
A July 13 survey from outplacement and career development firm Careerminds gives a sense of scale. The study polled 600 HR leaders who had overseen layoffs in the previous year. It found that 91.6% regretted the AI-related restructuring, while only 8.4% said the outcome matched expectations.
- 35.6% had already rehired more than half of the positions that were cut
- 52.1% brought people back within six months
- 30.9% said rehiring cost more than the savings from layoffs
- 42.4% said the financial result was roughly break-even
- 32.9% said critical skills and expertise had been lost
The report says that, taken together, about 70% of companies did not end up saving money. Data from staffing firm Robert Half adds another marker: 32% of U.S. hiring managers said they had eliminated a role because of AI and later reopened the same or a similar position.
Forrester expects more than half of AI layoffs to be reversed
Research firm Forrester said in “Predictions 2026: The Future of Work” that more than half of layoffs attributed to AI will be quietly reversed.
It also projected that by 2030, around 6% of U.S. jobs, or 10.4 million positions, will actually be replaced by automation, while another 20% will be augmented by AI rather than eliminated. J. P. Gownder, vice president and principal analyst at Forrester, said AI should be treated as a tool that amplifies labor, not a substitute for it.
Crypto companies are making similar AI claims
The report closes by noting that the crypto industry is no stranger to AI-led workforce narratives. Crypto.com CEO Kris Marszalek said last year, during a 12% layoff round, that people who could not adapt to AI would have to leave.
Coinbase said this month that more than 95% of its code is already written by AI, and estimated that by 2030, agent output would equal the workload of 100,000 employees.