The fear of AI job cuts keeps outrunning the payroll

People are expecting AI job cuts that aren't happening yet, at least not at the same level as expected...

A year ago, 32% of the people in McKinsey's state of AI survey (1,719 respondents from 97 countries) expected AI to cut jobs at their organisation. Over the year since, only 14% have actually seen it happen.

Now 39% expect AI-driven job cuts in the coming year, while two thirds say AI made little or no difference to their headcount over the last one. So nearly 70% said little to no difference... isn't that the real story here!

Why I think it happens

I often see headcount reductions put as the first line in the AI project business case, but it's one of the hardest benefits to actually realise as an outcome. Taking roles out means changing the work around them, the processes, the hand-offs and who owns each decision, and most organisations have not done that yet from what I see. McKinsey's own numbers back me up: outside the small group getting real value from AI, only a quarter have fundamentally redesigned their workflows. The tools arrived, the work stayed the same, so the jobs stayed too.

Two things to hold while you read it

McKinsey checked the comparison against the 552 people who answered in both years and got the same result, so this is not a quirk of the sample. If you have a headcount saving in your AI plan, it is probably not going to happen at the speed you are forecasting. And if you are the one worried about your role, the evidence says the fear is running well ahead of the cuts.

The action I would take

Find the headcount number in your AI business case and ask what has to change in the work for it to come true. If nobody can answer that, it is a benefits dream rather than a likely saving.