With regrets: CEOs who fired employees for AI are calling them back
Giant companies rushed to cut staff in the blind belief that artificial intelligence was ready to fully replace humans - but the awakening was painful: data from Forrester and Gartner, alongside operational crises at Ford and IBM, reveal that rapid automation has created engineering bugs, a collapse in customer service, and the erasure of organizational knowledge. This is the most expensive management U-turn of the decade.
In the last two years, many executives on Wall Street and in Silicon Valley were sure they had cracked the perfect economic model. The idea was simple and attractive to investors: cut significant percentages of the workforce in operations, customer service, and coding departments, declare a transition to artificial intelligence, and enjoy an immediate jump in the company's stock. This trend caused waves and reached a dramatic peak when data from the company Challenger, Gray & Christmas indicated that by mid-2026, tens of thousands of layoffs were recorded in the United States directly attributed to automation considerations. However, the reality on the ground was disappointing, frustrating, and above all, expensive.
The illusion of automation meets economic reality
A wave of global organizational regret, which could be dubbed the "Great AI Re-hiring," proves that the expectation that technology would fully replace humans was premature and too shallow. Companies are discovering that they did not save money, but instead created a deep operational crisis. Dry statistical data reveals the depth of this organizational fracture.
The latest trends report from the global research firm Forrester reveals that more than half of employers who carried out AI-based layoffs regret the move. CEOs rushed to eliminate positions even before their technological systems were mature or stable enough, and the results arrived quickly in the form of customer loss and damage to the product's core.
Operational U-turn and the realization: the machine needs a manager
Parallel to the regret data, a comprehensive survey by the global recruitment firm Robert Half found that nearly a third of organizations in the United States that eliminated positions due to automation implementation have already reopened the slots and hired human employees for the exact same roles. Senior managers explain that companies are forced to bring employees back to manage and fix tasks they thought the software would solve on its own in the real world.
Capabilities like judgment, oversight of edge cases, ethical thinking, and understanding the broad organizational context remain a distinct human asset. Technology, as of now, is simply incapable of replicating this intuition, and when it operates without close human supervision, its mistakes turn into a financial crisis.
To understand what this mistake looks like in practice, it is worth examining what is happening in the industry giants that managed these moves transparently. The automaker Ford has relied in recent years on automated systems and AI cameras to detect manufacturing and engineering defects, while allowing veteran engineers to leave the ranks.
The system missed critical malfunctions in real-time, which led to a sharp and sudden increase in expenses for recalls and expensive liability lawsuits. Ford's management admitted in interviews with Bloomberg that the company mistakenly assumed that the mere implementation of AI would produce a high-quality product, and forgot that software is only as good as the data used to train it.
The company was forced to re-hire hundreds of veteran and experienced engineers to oversee the systems and re-train them — a step thanks to which Ford managed to stabilize its quality metrics and save hundreds of millions of dollars in damages.
When dry rules encounter human complexity
Parallel to the engineering crisis, the technology giant IBM experienced a challenge of a completely different kind when it moved its human resources system to AI-based automation. The automated system successfully handled the vast majority of routine requests, such as issuing certificates or checking vacation days. However, a few percent of the requests — those that included complex ethical dilemmas, managerial sensitivities, or sensitive labor disputes — created a systemic paralysis.
It turned out that AI knows how to work according to dry rules, but it is completely devoid of emotional intelligence. Consequently, IBM's management changed course and announced a tripling of junior employee recruitment in the HR field in the United States. The heads of the company's human resources department clarified at professional conferences in New York that giving up on entry-level employees creates an existential danger in the long term, as without investment in the younger generation, the organization's talent pool will dry up and there will be no one to grow into senior management.
Another example of the limitations of replacing employees with automation too quickly came from the Commonwealth Bank, Australia's largest bank, which replaced dozens of human customer service representatives with an advanced voice bot. The bot failed to deal with complex financial requests from customers, which caused the opposite result: wait times skyrocketed, many calls were disconnected, and the volume of requests only increased. The bank was forced to cancel the layoffs, return the employees to their roles, and issue an official and embarrassing statement that the move was carried out without properly considering all the required business and operational considerations.
This case well illustrates the "economic glass ceiling" pointed out by Prof. Daron Acemoglu from MIT. From his calculations, a much more moderate picture emerges regarding the ability to replace employees using AI: according to him, only about 5% of tasks in the economy are expected to be candidates for automation via AI in a way that would be economically viable in the near future — a significant gap between what technology is capable of performing theoretically and what it is actually profitable for companies to replace.
Hidden costs and the collapse of the talent pyramid
Beyond the immediate damage to the quality of service and product, economists are now warning of two deep strategic dangers that many companies ignored during the panicked rush to automation. The first danger is the hidden computing costs. Many executives assumed that software is a cheap product that does not require maintenance, but they are discovering that the costs of processing, energy consumption, license acquisition, and fixing the bugs that AI produces sometimes cost more than the general salary of the employees who were fired. The machine does not operate for free, and managing its errors requires vast resources.
Another danger is the collapse of the traditional workforce structure and the transition to a diamond model in the labor market. In the past, companies were built in a pyramid structure: they hired starting and cheap employees, and they grew within the organization and became experts who know the company's DNA. Automation erased the junior layer, but now companies are discovering that they have no one to oversee the AI, because there are no young employees acquiring basic experience in the field. The result is a rigid and unprecedented demand for expert and experienced employees, whose price in the market is rising and forcing companies to pay particularly high salary premiums to bring them back to the company after they were fired.
Warning sign for the Israeli manager: productivity is not replacement
In recent times, the global labor market is undergoing a healthy process of sobering up and re-evaluation. Research companies like Gartner are already estimating that by 2027, half of the companies that reduced their workforce in customer service and attributed the move to AI will re-hire employees to perform similar roles, even if under new job definitions.
According to a previous forecast by Gartner from June 2025, by 2027 half of the organizations that planned to significantly reduce their workforce in customer service will abandon these plans. These data are a critical warning sign also for the Israeli economy and high-tech, which in the last two years have rushed to copy the aggressive layoff models from Silicon Valley in the name of efficiency.
The necessary conclusion is that artificial intelligence is an exceptional work tool capable of dramatically improving the productivity of the existing employee, but it is a poor, rigid, and dangerous human replacement. The companies that will win in the next decade are not those that will rush to cancel jobs to show temporary savings in quarterly balance sheets, but those that will know how to combine the two, out of a deep understanding that human capital, accumulated experience, and the direct connection with reality are still the most stable and profitable economic asset of the organization.
