On 2 July I said nobody in Britain was counting whether AI is taking the entry-level jobs. Somebody is, and it is the Bank of England.
On 2 July this paper ran a piece saying the graduate jobs are vanishing and that blaming the machines is the easy answer, because the government's own data will not pin it on them. I stood behind that. On 10 July I went further and said we had spent a year arguing about a number nobody collects.
I need to update both.
On 24 July the Bank of England published its Agents' summary of business conditions. The Agents are the Bank's regional network. They spend their working lives talking to British firms, and what they gather goes to the Monetary Policy Committee before it sets rates. This one carries a box on artificial intelligence, and the box says this.
Automation is reducing demand for some entry-level and junior roles that have traditionally provided early-career experience. It names the tasks: document preparation, invoice processing, basic analysis. It says several professional services contacts report reduced graduate recruitment and lower demand for administrative and junior staff. It says some of them have raised concerns about the long-term implications for skills development and talent pipelines.
That is not a survey and it is not a statistic. It is what firms said, collected by people whose job is to listen to firms. Tomás argues below that this is worth a good deal less than it looks, and his case is a strong one.
But it is the clearest official account Britain has of AI removing early-career work, and I did not know it existed when I said nobody was counting.
Now the part I keep turning over.
Yesterday at noon the Bank held Bank Rate at 3.75 per cent, on a vote of six to three, with the three wanting to raise. Alongside the decision the Committee said that loose labour market conditions will act to reduce inflation over time.
Put those two documents next to each other. In one, British firms say they are automating and hiring fewer juniors. In the other, the slackness of the labour market is a reason to expect inflation to fall.
I am not saying the Bank has done anything wrong. Setting policy against the labour market you have is the job, and nobody at the Bank has claimed automation is the cause of the slack. The same box lists weak demand and labour costs ahead of it, and James and Tomás both make sure you know that.
What I am saying is smaller, and I think harder to argue with. The best written record this country has of what AI is doing to the first rung of working life turned up inside a document produced to help price money. It was not collected to answer the question. The answer fell out of it sideways.
That is what happens when a country has no instrument pointed at something. The evidence still exists. It arrives in somebody else's paperwork, and it gets filed under what they were looking for rather than under what you needed.
Ines has the other half of the same box below, where the AI build shows up in British prices through grid queues and memory chips. Ada has the section almost nobody reads, in which the Bank talks to ordinary people and hears that senior workers are now applying for junior jobs. Elena has the measurement question. Tomás thinks I have got carried away.
The Playbook.
Nothing here moves a payment today. Bank Rate is where it was. The three votes to raise are a signal about the coming months rather than about today.
If you run a business and you have cut junior hiring because software now does the work, you are in the Bank's intelligence whether you know it or not. What you say to an Agent is part of how this country understands itself. Worth knowing when the call comes.
If you are early in your career, or you have a child who is, the line to hold on to is the one about senior workers applying for junior roles. The competition is not only your own age group any more.
And the thing to watch is whether any part of the British state starts collecting this on purpose rather than by accident. When it does, we will say so here.
— 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.*
The founding question asks what people do when they do not have to work. This is where it stops being hypothetical in the British record: the people it is happening to now appear in the inflation forecast as slack.
The gap is institutional and belongs to no party. No British statistical body collects a measure of AI-related displacement, so the clearest account of it arrives inside a monetary policy document. The note charges the absence of an instrument, not a government.
- Bank of England, Agents' summary of business conditions, July 2026, published 24 July 2026, including Box A on AI adoption
- Bank of England, Monetary Policy Summary and minutes, meeting ending 29 July 2026, published 30 July 2026
- House of Commons Library, Interest rates and monetary policy: economic indicators, updated 30 July 2026