Kurt Gödel left 1,009 pages of dense mathematical notes from the early 1940s, much of it written in Gabelsberger — a German shorthand system that was already obsolete when he used it. A small team at the University of Helsinki, led by logician Jan von Plato, had been working through them for years at roughly one page a week.

Then, at a conference on AI and theorem proving in France in August, colleagues suggested he hand the material to a model. Von Plato gave Claude his own transcriptions of the workbooks, together with the six books he had written about Gödel. "In a couple of days, I got to see a 650-page explanation of the workbooks," he told The Guardian, which reported the work on Friday. "It had to be controlled page for page, but it's basically a work that is impossible to do for human beings."

The useful output was a catalogue: 663 theorems for which Gödel had a proof, a sketch of one, or at least an idea for one. Von Plato ran them through Coq, a proof assistant that mechanically checks whether an argument is valid. About 400 were verified as stated, roughly 200 could be repaired by tightening a statement or adding an assumption, and about 50 were false. "I was flabbergasted," he said.

The most striking find concerns an old dispute. Paul Cohen won the 1966 Fields Medal for showing that the axiom of choice and the continuum hypothesis are independent of standard set theory. Gödel insisted he had proved the same thing in the 1940s, but when colleagues pressed him he could not reconstruct the proof. Claude surfaced an almost complete argument in the notebooks — one definition short of finishing, and Gödel had written the correct definition two years earlier in another volume. Von Plato put the repaired proof through Coq and it passed.

Outside review has not happened yet. Joel David Hamkins of the University of Notre Dame told The Guardian that if the reconstruction holds it is "remarkable and historically interesting", but added: "How much is Claude and how much is Gödel is a job for the historians." Unpublished work, he noted, does not earn credit. Von Plato plans to continue: he says the notebooks hold more than a dozen logic and foundations themes nobody has pursued since, and Gödel's unstudied physics and cosmology notebooks are next.