The 38,000 does not count drivers. It counts the people who build the thing that replaces them

Open the document the number comes from and the sentence is precise in a way the repetitions are not.

Connected and Automated Mobility 2025, command paper 719, presented to Parliament in August 2022. The relevant line says the UK alone could be worth as much as forty-two billion pounds by 2035, capturing around six per cent of a six hundred and fifty billion pound global market, and creating up to thirty-eight thousand new jobs in the sector.

In the sector. Three words, and they carry the whole thing.

The sector means the industry that builds and runs this technology. Engineers writing the software. Fleet technicians maintaining the vehicles. Insurers pricing a risk nobody has priced before. Certifiers, regulators, the people who install and calibrate sensors, the people who staff the remote operations rooms.

It is a count of an industry being constructed. It is not a net employment figure, it does not subtract anything, and it does not count drivers at all.

That distinction survives in the command paper and disappears almost everywhere else. By the time the figure reaches a headline in June 2025, it is thirty-eight thousand jobs on the horizon, and a reader has no way of knowing that the jobs on that horizon are not the jobs anybody is worried about.

The same command paper also says what it thinks the labour problem is. It names the shortage of heavy goods vehicle drivers as its worked example of a skills challenge, and offers self-driving technology as one response. Not as a way of managing too many drivers. As a way of coping with too few people willing to do the job.

The paper is careful about even that. It says the technology is by no means a panacea.

So the number was built in 2022, for a country that could not find enough drivers, to describe an industry that would need building. It is now being quoted into a very different argument, about what happens to drivers who have the work and might lose it.

A number does not become wrong by being reused. It becomes uninformative, which is harder to notice.

There are two figures in circulation about self-driving vehicles and British employment. The government's thirty-eight thousand, in the sector, by 2035. And a union's three hundred thousand driver jobs at risk, agreed at a congress in June.

They are a factor of eight apart, and they are not opposites. One counts jobs in the industry that builds the machines. The other counts jobs in the trade the machines would displace. They measure different populations, on different timescales, and they were never designed to be compared.

Somebody could reconcile them. It would require deciding what counts as a job created, what counts as one displaced, over what period, and in which parts of the country. That is a piece of analytical work, not a political statement, and the International Transport Forum, an OECD body Britain belongs to, has twice recommended exactly this kind of anticipatory approach to transport workforce transition, in 2023 and again in April 2026.

In four years, nobody has been asked to do the subtraction. Both numbers keep being quoted, at each other, by people who are not describing the same thing.

One loose end belongs on the record rather than tidied away. In the original command paper, the thirty-eight thousand carries a footnote. I have not read it. Neither, as far as I can establish, has anybody who has quoted the figure in the four years since. It is the provenance of the most repeated employment number in British transport policy and it is one page turn away.