Is AI Really Gutting the Workforce Yet?

Open any feed and you will find the same warning: artificial intelligence is coming for the jobs, and the layoffs have already begun. It is one of the most widely held fears about the technology, and it feels intuitively right when a company announces cuts and mentions AI in the same breath. The trouble is that the best evidence we have so far does not show a workforce being hollowed out. It shows very little happening at all.

The clearest look comes from a working paper circulated through the National Bureau of Economic Research, which surveyed nearly 6,000 chief executives, finance chiefs, and senior managers across the United States, the United Kingdom, Germany, and Australia. More than 90 percent of the firms reported no measurable effect on their employment from AI over the previous three years. Around 89 percent reported no measurable change in labour productivity either. This is not a study by AI skeptics. It is companies describing their own results, and most of them say the needle has barely moved.

Adoption is real, impact is not

It is not that nobody is using the tools. About 69 percent of the surveyed firms actively use AI in some form, and more than two thirds of the executives use it themselves. But that usage averages roughly an hour and a half per week, which is closer to occasional help than to a machine quietly replacing a department. Widespread dabbling, not wholesale transformation, is the honest description of where most companies sit.

The researchers also flagged something they call a productivity paradox: the gains people feel they are getting are larger than the gains that actually show up in the numbers. That gap is worth sitting with. It may mean the benefits are real but lag behind, taking time to turn into revenue. It may also mean some of the enthusiasm is just enthusiasm.

So where do the layoffs come from?

Here is the uncomfortable part for both sides of the argument. Companies are cutting jobs, and many of them cite AI when they do. Yet the same executive class, asked in detail, largely reports no productivity boost from the technology so far. That tension suggests AI is sometimes a convenient label for decisions made for older reasons, such as over-hiring during the boom years or a slowing economy, rather than the sole cause of the cuts.

None of this means the fear is baseless, and it would be a mistake to read the survey as an all-clear. The same executives expect change ahead. On average they forecast that over the next three years AI will lift labour productivity by around 1.4 percent and trim employment by about 0.7 percent, with roughly two thirds of that workforce reduction coming not from firing people but from hiring fewer of them. A slow squeeze on new openings is easy to miss and hard on anyone entering the job market, even if it never produces a dramatic wave of redundancies.

The honest reading

The claim that AI is already gutting the workforce runs well ahead of the evidence. What the data supports is narrower and less dramatic: broad adoption, modest measured impact, and a real but gradual effect that shows up first in the jobs that are never advertised. The picture could change quickly, and surveys of executive opinion have obvious limits. For now, though, the mass extinction of jobs is a forecast people keep repeating as if it were a headline that already ran.