The Credibility Gap: AI Companies Are Selling a Story Their Own Structures Don't Believe

The Lede

When Nike announced roughly 1,400 job cuts last year, it said the restructuring was partly about freeing up resources for AI. When Duolingo cut its translation team, it said AI could handle the work. When a dozen other companies explained their layoffs to investors and employees, they reached for the same phrase.

The only problem: the AI did not exist.

According to Forrester analyst J.P. Gownder, when his team asks companies that have announced AI-driven layoffs whether they have a mature, vetted AI application ready to fill those roles, nine out of 10 times the answer is no. They had not even started building it.

Nine out of 10. That is not a rounding error or a bad quarter. It is an industry telling on itself.

The Dirty Secret

The framing that AI is eating jobs has become the default explanation for corporate restructuring. Workers laid off in 2025 heard it in the internal memo. Investors heard it on the earnings call. The press release reached for the same word: AI.

But the data tells a different story. Challenger Gray Christmas, which tracks US layoffs, found that AI was cited in roughly 55,000 job cut announcements in 2025. That sounds like a wave. It represents 4.5 percent of the 1.2 million total layoffs last year. AI was the stated reason for one in 22 cuts. The other 21 had nothing to do with it.

Of the companies that did cite AI, most did not have the system running. A February 2026 Gartner survey of 321 customer service and support leaders found that only 20 percent had actually reduced agent staffing because of AI. The other 80 percent were still carrying the same headcount while the press release used different language.

And they know it. Gownder calls it AI washing: attributing financially driven cuts to a future AI capability that has not arrived. It is a convenient cover story for decisions that would have been made regardless.

The Regret Rate

The companies that went ahead and cut are now having second thoughts. According to Forrester, 55 percent of employers report regretting laying off workers for AI. They made the announcement, took the cost savings, and discovered the work still needed to be done.

Gartner's prediction is more specific. By 2027, the firm forecasts, half of companies that attributed headcount reductions to AI will quietly rehire for similar roles. Not the same jobs, exactly — the titles will be different, the team structures tweaked. But the work will be back, performed by people who were told a machine had made their position unnecessary.

This is not a market correction. It is an admission. The industry predicted AI would replace 6.1 percent of US jobs by 2030, roughly 10.4 million roles. What it did not say is that nine out of 10 companies would try to make those cuts before the AI existed.

What AI Actually Did

The jobs that were actually eliminated tell a narrower story. Junior roles, customer service positions, and some software development work have seen real pressure. But even in customer service — the category most frequently cited as ripe for AI replacement — the survey data shows most leaders have not reduced headcount. They are experimenting with AI tools alongside their human staff, not swapping them one for the other.

The number that matters here is not the 10.4 million roles forecast to disappear. It is the 20 percent of jobs that Gownder says AI will reshape or augment. More jobs will be changed by AI than replaced by it. The ratio is roughly three to one. Companies are using the replacement framing to justify cuts that are really about margins, restructuring, and investor expectations.

The version of AI that would justify mass displacement does not exist in production at the companies that announced the layoffs. It exists in the pitch deck.

What They Built Instead

There is one thing the AI is definitely doing inside these companies: watching.

Marc Benioff, the Salesforce CEO, said on the All-In podcast that he uses Slack's AI to query what employees are upset about. He can ask what the top five deals are, what the top complaints are, what needs his attention. The AI reads the direct messages and channels. It is not replacing the sales team. It is monitoring them.

This is the actual AI deployment at companies announcing AI-driven layoffs. Not the work, but the surveillance. The technology that exists today is good enough to watch what people are writing, flagging what deserves attention, summarizing what happened. It is not good enough to do the jobs those people were hired to do.

The Boiling Frog

There is an old line about AI being too cheap to meter, a phrase borrowed from the nuclear industry's most famous overpromise. The comparison is not accidental. Both industries built their public narratives on a future capability that their own internal structures treated with considerably more caution.

The difference is that the nuclear industry eventually had to answer for the gap. The AI industry is still in the window where the gap can be reframed as a timing problem rather than a credibility problem.

That window is closing.

The companies that announced AI-driven layoffs in 2024 and 2025 are now in the phase where the work still needs to be done, the AI has not arrived, and the cost savings are not materializing the way the model projected. The 55 percent regret rate is the floor, not the ceiling. The reversals Gartner predicts for 2027 are already starting to appear in scattered hiring data.

The story the industry is selling — that AI is replacing human work at scale — is not false because AI will never get there. It is false because it is happening now, before the technology is ready, and the people who said otherwise are now on record.

The Short Version

The AI that was supposed to replace workers does not exist at the companies that announced the layoffs. The workers are gone anyway. The companies know it. More than half of them already regret it. Half will rehire by 2027. That is the story.