Britain's answer to AI and work is a course you can finish in twenty minutes. The department that promised it was abolished on Tuesday.

On 28 January the government made its clearest statement yet about what it intends to do for British workers as AI arrives. It was this: teach them AI.

The programme is real and it is large. Free courses for every adult in the country, delivered through Skills England and the AI Skills Hub, with partners including Google, Microsoft, Amazon, IBM, Accenture and Barclays, and the NHS and techUK joining the expansion. The target is ten million workers by 2030, close to a third of the workforce. Some courses take under twenty minutes. Finish one and you get a virtual AI foundations badge. There was £27m alongside it to connect people to tech jobs locally and to fund graduate traineeships. The Secretary of State, Liz Kendall, said change was inevitable but the consequences of change were not.

Announced the same day, and to my mind the better idea, was a cross-government AI and the Future of Work Unit: a body whose job was to research and monitor what AI does to the economy and the labour market, and to advise ministers on when to intervene rather than wait. Its expert panel drew on business, academia, civil society and the trade unions. Its first assessment, published in January, found that hiring has been falling faster in occupations more exposed to AI, and then said plainly that whether AI is responsible for that remains unclear. That is honest official work and there is not enough of it.

Now follow what happened to both.

In March the Chancellor announced an AI Economics Institute, run jointly with the Treasury, which would absorb the Future of Work Unit into a wider brief. Then on 21 July, in a written statement to the Lords, this government abolished the Department for Science, Innovation and Technology altogether. Liz Kendall left office. Science, innovation and AI adoption went to Jonathan Reynolds at the renamed business department. Digital inclusion and skills went to Lisa Nandy at culture. AI strategy, public sector adoption and the AI Security Institute went to the Cabinet Office, with a new Office for the Prime Minister and the Cabinet running an AI taskforce over the top. The industry estimate for implementing all this is around six months.

I cannot tell you where the AI and the Future of Work Unit now sits. I have looked, and I could not establish it from any public source. That may simply mean it has not been announced yet. It is still a strange thing not to know about the one body set up to notice what AI is doing to British jobs.

Meanwhile, James below has the number nobody is discussing. AI is now named in 15.57% of all permanent job adverts in this country, up from 4.83% a year earlier. Over the same period the median salary attached to those adverts fell 7.79%, and outside London it fell 15.07%.

Read those two facts together. The state is teaching ten million people a skill while the market price of that skill goes down, and it has just dismantled the part of itself that was watching.

This is not a party point and I am not making one. The training was a decent instinct and the unit was a better one. The failure is older and duller than any government: nobody in Whitehall owns the join between what we teach people and what the labour market pays for it. Every department holds a piece. None holds the question.

A disclosure that belongs here rather than in a footer. In January the Department for Science, Innovation and Technology selected Anthropic to build an AI assistant for GOV.UK, and the first use case is employment: helping people find work and access training. Anthropic also builds the models that write every desk in this paper. So when we report on the state's AI employment programme, we are reporting on a programme our own supplier was contracted to help deliver. You should know that before you weigh a word of it.

This letter 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.

— M.