The AI handled 60% of the job. Turns out that was the wrong 60%.

The playbook, apparently, goes like this: announce AI is handling it, downsize the team, watch AI complete 60% of the job while completely blanking on the other 40%, then post the same listing six to twelve months later. Time reported on the pattern this week, and the numbers are there to back it up.

Robert Half found that 29% of companies that cut staff for AI have already hired back into the same positions. Forrester predicts 55% of executives who swapped humans for AI will regret it within 18 months. We are currently inside that 18-month window. The regret is showing up in job postings.

What went wrong? AI is good at tasks. It is not good at jobs. Customer service reps don't just answer questions — they know when someone is actually upset versus just venting, when a high-value client has been on hold one too many times, when something needs a supervisor. Marketing copywriters have a "voice." Bookkeepers catch anomalies on instinct. Sales teams close deals through relationships. Harvard economist David Deming found AI doesn't replace social skills or judgment — it makes them more valuable.

The math doesn't work either. The employee laid off at $55,000 is coming back at $75,000 — because now they also manage the AI that took their job. Add recruitment costs, onboarding, and the months of institutional memory you lost, and the savings are gone. Gartner forecasts that by 2027, half the companies that cut customer service jobs for AI will be rehiring into the same roles. But the original employees have moved on. The replacement costs more. And the team that stayed watched the whole thing happen.