Ford has spent three years rehiring roughly 350 veteran engineers, in the company's own phrase with "gray beards",  many of them former staff, to repair quality failures that automated inspection could not. Their brief is the tell: not to replace the tools that fell short, but to fix them, and to train the juniors those tools were meant to make unnecessary. Ford's own hardware chief admitted the misjudgment plainly, the assumption that introducing artificial intelligence would, by itself, yield a quality product. It did not.

IBM automated much of its HR function, found the system could not hold judgment or the off-script case, and has now pledged to triple its US entry-level hiring. Commonwealth Bank of Australia replaced service staff with a voice bot, watched call volumes climb, and reversed the redundancies.

The pattern beneath the headlines is not that the technology cannot work. It is that two things broke quietly on the way to proving it. The first is tacit knowledge, the judgment that tells an experienced engineer which anomaly to flag and which to wave through, drawn from a hundred edge cases and recorded in no manual.

The second is the apprenticeship pipeline: automate the entry-level work and juniors have nothing to learn from, while the veterans who would have taught them have already left. IBM's own HR chief named the risk, no pipeline, and the well simply dries up.

For anyone who felt most disposable through the AI panic, the market has issued a correction. Experience, judgment and the slow accrual of pattern, the very things that never appeared on a process map, have become even more valuable.

"AI replaced them" was never the whole sentence. The second half is arriving now, with a price attached for the companies who fired too soon.