Over Half of Leaders Who Cut Jobs for AI Say They Got It Wrong. They Cut Capacity Before They Understood the Work.

TLDR: More than half of leaders who made AI redundancies say they got them wrong; they cut the roles before mapping the work, so headcount should move last.

A bank called its own voice-bot redundancies an error

In August 2025, Commonwealth Bank of Australia (CBA) reversed the redundancy of 45 customer-service roles it had tied to a new voice bot. The explanation the bank gave was unusually candid. Its initial assessment that the roles were not required “did not adequately consider all relevant business considerations”¹, and the bank called that an error. The bank framed the failure as an incomplete assessment of the work, where a forecast miss would have pointed at the bot.

That distinction is the subject of this article. The most quoted statistic on artificial intelligence (AI) layoffs says that more than half of leaders regret them. Read carefully, the data says something narrower and more useful. And the cases behind it describe the same mistake again and again: leaders priced a task, removed a role, and discovered afterwards what else the role had been doing.