Somewhere in the last two years, a lot of businesses made a bet. They looked at what AI was expected to be able to do, not what it could actually do yet, and cut roles accordingly. Forrester's research on 2026 workforce trends has a number for how that bet is playing out: 55% of employers now regret AI-related layoffs. More than a third have already quietly rehired over half the roles they eliminated, often within six months of making the cut.

The word researchers have settled on for this is the boomerang, and it's a useful one. It captures both the shape of the story and the reason it happened. These weren't considered decisions based on what AI was actually delivering inside a specific business. They were decisions based on what AI was expected to deliver, in general, sometime soon. The gap between those two things turned out to be expensive, and it's the people who were let go, then quietly asked back, who paid for it first.

Why the gap opened up

It's worth being precise about what went wrong, because "they moved too fast" undersells it. The actual mistake was treating a general capability claim as if it were a specific, tested fact about one business's own work. AI genuinely can do a great deal. Whether it can do the specific, judgement-heavy parts of a specific role, reliably, without a human catching the errors, is a different question, and it's one that only gets answered by looking closely at the actual work, not the press coverage about the technology in general.

Businesses that skipped that step made a decision that felt bold and turned out to be closer to a guess. The rehiring wave is the market quietly correcting for that guess, role by role, once the theoretical case met the reality of the work not getting done properly without a person there to do the parts AI genuinely couldn't.

The professionals who didn't get cut

There's a second half to this story that's easy to miss, because it's less dramatic than a layoff headline. The people who kept their seats, in businesses that didn't make this mistake, tended to be the ones who had already worked out what to hand to AI and what to keep for themselves. Not through luck, but because someone had actually looked at the work closely enough to know the difference.

That's the distinction worth sitting with. It isn't AI skill in the abstract, and it isn't resistance to AI either. It's specific, tested knowledge of where a tool genuinely helps in a particular role and where it doesn't, held by someone who took the time to find out before making a decision that affected a person's job.

What this means for a smaller business making its own AI decisions

Most SMEs aren't making headline-grabbing layoff announcements. But the same underlying mistake, acting on what AI is generally said to do rather than what it's actually shown to do in your specific business, shows up in smaller ways constantly: a role reshaped around an assumption that hasn't been tested, a hiring decision paused because "AI will probably handle that soon," a budget line cut because a vendor's pitch deck was persuasive.

The fix isn't caution for its own sake. It's the same half-day of honest groundwork that would have stopped the boomerang before it started: looking specifically at where AI actually changes the work in your business, tested against your real operations rather than a general claim about the technology, before any decision gets made that's hard to reverse.

Going wisely was never really a slogan. It's what the data now shows was the better strategy all along, and the businesses currently rehiring the people they let go are the clearest evidence of that.

Go wisely.

Source: Forrester Predictions 2026, reported via HR Executive and other outlets, March-April 2026.