Two institutions answered our questions on Monday. One of them told us where the hour goes

Yesterday in this slot I wrote that when work gets faster the saving has to land somewhere, and that almost nobody in British public life is asking where. I named three possible destinations: the customer, the worker, the owner.

On Monday two answers arrived, both to questions this paper had put in writing and waited on.

The first is the one that matters.

In September Deloitte published the largest survey of workplace AI use ever done in one country. It said that workers save about seventy minutes a week using these tools, and that most of the time saved goes into doing more work for the same employer. It did not publish the question it had asked, or the number behind the word most. On 21 September we asked for both.

On Monday afternoon it sent them. Of the workers who said the tools saved them time, 84 per cent said they used that time to do more of the same work, or different work, for the same employer. Twenty-six per cent said they were doing anything else at all.

So there is the answer, from a named source, on the record. Roughly five in six of the workers who gain an hour give the hour back. Not as a shorter day. Not as pay. As more work.

Set that beside the other thing Deloitte told us. Of the people buying these tools out of their own pocket, 83 per cent are employees, the group that on HMRC's test will usually get no tax relief on the cost. The worker buys the tool from taxed pay. The employer collects the time. James has both answers and what they add up to.

I should say what we have not done with this. Deloitte's account of how it reached its £958 million figure does not sit easily with the wording of its own press release. That is a serious thing to say about a firm's published number, so we have put it to them and we are waiting. It is not in today's paper as a finding.

The second answer came from the Office for National Statistics, to four questions we sent on 21 September about whether its own survey has been skewing the jobs figures. One answer was that it has made no assessment. Then, on the same morning, it published research putting a number on the drift: the survey undercounted employees by 365,000 a quarter between 2022 and 2024. Leah has it.

Two institutions, on one Monday, showing their working because somebody asked. This paper has run the line that nobody is counting often enough that our own Contrarian has taken us to task for it twice in a fortnight. He was right, and Monday is the evidence. Asking is not the same as nobody counting, and we should be more careful about the difference.

One other thing, since the clock is running. The government's consultation on monitoring at work closes at 11:59pm on Wednesday. We pointed readers at it a week ago. If you meant to respond and have not, the Playbook has the routes.

— M.

This note is mine: the view, and the call to run it. It begins as a draft, drawn from work the AI and I have researched and argued out together, the same way every desk in this paper is made, and I answer for every line because I read every line. Those desks run on models built by Anthropic, one of the labs sitting on the very scoreboard we report, so we tell you plainly: we cover this from inside it.