What comes to mind, reading these sentences? “The challenge is not technical but political, requiring careful sequencing, clear communication, and sustained commitment.” Or: “Private ownership is not rewarding innovation but collecting a toll on Victorian infrastructure while deferring the investment required to keep it functional. The technology has not changed. The ownership has—and the ownership is the problem.”
Boredom first, probably. Policy writing can be sandpaper-dry. But something else might stand out too, at least to anyone who often interacts with artificial-intelligence models like ChatGPT or Claude. Prose produced by those AI tools has a distinctive smell. One telltale whiff is a love of contrasts: “not technical but political”; “The technology has not changed. The ownership has”. Another is tricolons. Humans have long appreciated groups of three (Cicero studded his speeches with them), but AI’s enthusiasm is superhuman: “careful sequencing, clear communication, and sustained commitment”. And then there are the em-dashes—everywhere.
Readers today can do better than guesswork and amateur stylometry, though. Over the past few years, AI detection technology has improved dramatically. These tools use similar machine-learning techniques to AI models. Pangram, a market leader, boasts of a 0.01% false-positive rate. A study by researchers at the University of Chicago found it had a “near-zero” error rate on longer texts.
Pangram flags each of the policy papers those quotes came from as being mostly written by AI. One was by Louise Haigh, an MP and close adviser to Andy Burnham, Britain’s prime-minister-in-waiting. The other was “The Productive State”, a manifesto of nearly 70 pages published by Mainstream, a Labour group linked to Mr Burnham; it offers a “Framework for Manchesterism”, Mr Burnham’s still-loosely-defined economic programme. Ms Haigh said, in a statement to the Telegraph, that AI was used for “version control”. Mainstream told The Economist that one co-author used Claude to “expose line-level differences and unify the voice of both authors’ independently drafted sections through copy editing”.
That raises two questions. Is this AI writing a problem? And why the links to Mr Burnham? Start with the first. Politics is full of prose not written by the apparent author: politicians have speechwriters, intellectuals employ research assistants. Perhaps AI is no different, even if the prose is clunkier. But people setting out ideas in politics are asking for something quite audacious: to reshape how a country is governed around what they think. Any writer knows how much putting words on a page can tighten one’s arguments. Even subcontracting that to an aide beats skipping the (sometimes painful) process entirely. AI certainly produces sloppy prose, but it also papers over sloppy thinking.
Maybe that matters less if, as Ms Haigh and the “Productive State” authors say, AI is used only for a late polish. The trouble is that readers can detect whether AI wrote the final product, but not how authors used it. And when AI’s involvement is disclosed only after the press come knocking, trust takes some earning back.
What about the potential Burnham factor? Mr Burnham’s coup was plotted slowly, but executed quickly. From the moment he won the Makerfield by-election, he was all but guaranteed to end up in Number 10. Yet he is still vague on many of the big questions facing Britain. That mismatch offers political entrepreneurs a chance to try to shape the wet clay of Burnhamism around their own priorities. The currency of ideas in Westminster is PDFs, ideally long ones. And when you need one quickly, who better to turn to than Claude?
For more expert analysis of the biggest stories in Britain, sign up to Blighty, our weekly subscriber-only newsletter.