AI layoffs count the jobs cut. Who counts the hours freed?

AI layoffs get most of the headlines: a company announces how many jobs AI will remove. The companies themselves report something smaller. In a survey of nearly 9,000 CFOs, CEOs and senior executives in seven countries, published by the National Bureau of Economic Research and revised in October 2026, executives reported "essentially no net effect" of AI on their firm's employment over the last three years. They put AI's contribution to labor productivity growth at 0.2% a year over the same period.

So far, AI has changed the number of jobs very little overall, and added little measurable output. Yet AI is being used. The same executives say they use it for about 2.1 hours a week themselves. If those hours save time, the 0.2% shows the saving is not reaching the output.

A headcount plan has nowhere to put the time

A plan built on job cuts has one measure: the number of roles removed. When AI saves a team a few hours a week and no role disappears, that plan records nothing. The hours still exist, spread across many people, and nobody has decided what they are for.

Without a decision, those hours go to whatever is closest. People check the AI's output more carefully, answer more messages, or pick up work that was waiting. Some of it is useful. But nobody chose it, so nobody measures it, and the saving does not show up in productivity or in the budget.

The executives in the survey expect AI's boost to labor productivity to rise to 0.6% a year over the next three years. In our view, that depends on the freed hours going somewhere on purpose.

Decide what the hours are for before the project starts

Before an AI project starts, the leadership team names what the released time will be used for: more time with clients, faster turnaround, training, or work the team has never had time to start. Then it measures that result, the same way it would measure a cost saving.

This is also where the team comes in. People who know what their freed hours are for will help find them. People who expect the saving to be counted in heads have every reason to keep quiet about it.

If no one decides what the freed hours are for, they disappear into busy work.

Three questions for your leadership team

1. For each AI project we run, have we written down what the hours it frees will be used for, and who owns that decision?
2. How will we measure what those hours produce, the same way we measure a cost saving?
3. Do our teams know whether the time they save will be counted as fewer roles or as new work?

Return on intelligence is one of Wyrework's services. We work with leadership teams to decide, before an AI project starts, what the hours it frees are for, and then we measure that result.