Britain just finished its biggest rethink of school in a generation. It never asked what school is now for.

Every so often a country stops and asks itself what its children should learn. Britain just did. The Curriculum and Assessment Review, led by Professor Becky Francis, ran for over a year, took evidence from thousands, and reported in November 2025 at nearly two hundred pages. It is the most thorough look at what English children are taught, and how they are tested, since the exams most of today's parents sat.

Read it and you find serious people doing serious work. Scrap the English Baccalaureate as a school measure. Cut some of the exam load. Add a proper framework for speaking well, not just writing well. Keep the curriculum, in its own words, knowledge-rich. Its chair described the whole exercise, more than once, as evolutionary not revolutionary. A revised curriculum will be published in 2027 and taught from 2028.

Now hold it up against the thing happening outside the classroom window. The machines this paper covers every day are getting good, fast, at precisely what school measures: recalling the fact, structuring the argument, writing it up under a word count. The review asks, carefully, how to teach and test that better. It does not ask the question underneath it: what is that skill worth, in the working life of a child who is five now and leaves school in 2039, if a free tool does it in seconds?

That question is not in its terms of reference. And it is not in anyone else's. Listen to how every party talks about education and you hear one word over and over: skills. The government's post-16 white paper, published in 2025, promised a world-leading skills system. Skills for what, and for how long, is the part that never gets said. The political class treats more education as a self-evident good, the way it once treated more motorway, and argues only about the amount and the bill.

This is the gap this strand exists to name, and it belongs to no single party. The review was commissioned under one government and will be delivered under another. Both framed it the same way. The failure is shared, and it is a failure of imagination rather than effort: everyone now accepts that AI is reshaping work, and no one has connected that to the thirteen years we spend preparing children for work. We are running the two biggest facts of the decade, that the machine can do knowledge work, and that we school children to do knowledge work, in two separate rooms, and letting neither see the other.

None of this makes the review wrong to fix what it fixes. A better test beats a worse one. But you can improve the measurement of a thing while missing that the thing itself is losing its value. Britain has just spent a year, and a great deal of expertise, making the exam hall run better. It did not spend an afternoon asking whether the exam hall is still pointed at the right target. For a country that sells knowledge for its living, that is the omission that should be keeping a minister awake. It is keeping no one awake at all.