Research read

Claude helps logician map 663 results in Kurt Gödel's notebooks

Helsinki logician Jan von Plato used Claude to digest 1,009 pages of Gödel's 1940s workbooks in days, surfacing 663 theorems and a near-complete proof Gödel claimed in 1966. It is one of the clearest examples yet of a language model accelerating archival scholarship, with formal verification as the check.

AE
Aidose Editorial
Nov 15, 2026

Jan von Plato, a Finnish logician who has led a project at the University of Helsinki on Kurt Gödel's unpublished notebooks since 2017, says Anthropic's Claude produced a comprehensive digest of 16 Gödel workbooks in a couple of days, work that had been progressing at about one page a week. The Guardian reported the results on November 13.

The workbooks, written in the early 1940s, run to 1,009 pages, much of it in Gabelsberger, a German shorthand that was already obsolete when Gödel used it. Claude did not read the shorthand itself. Von Plato gave it his own transcriptions of the workbooks, six books he has written on Gödel and other secondary literature, and received what he described as a 650-page explanation of the material, which he said "had to be controlled page for page."

Asked to list every theorem for which Gödel had a proof, a sketched proof or an idea for a proof, Claude returned 663. Von Plato told the Guardian that when the results were run through the Coq proof assistant, about 400 were verified as stated, about 200 needed repair, such as a more precise statement or an added assumption, and about 50 were false.

The first question von Plato put to Claude was whether the workbooks contained a proof of the independence of the axiom of choice, a result for which Paul Cohen won the 1966 Fields Medal and which Gödel later said he had proved in the 1940s. According to von Plato, Gödel was "one step away": one definition was wrong, though Gödel had written the correct one two years earlier. Von Plato said the reconstructed proof passed Coq. Joel David Hamkins of the University of Notre Dame, who has not seen the reconstruction, told the Guardian that if it holds up it is "remarkable and historically interesting," but that how much is Claude and how much is Gödel "is a job for the historians."

For researchers, the workflow is the notable part: a long-context model used to connect material across a thousand pages, with a formal proof assistant, not the model, serving as the arbiter of correctness. Von Plato says the workbooks contain more than a dozen unpursued themes in logic, and he plans to turn next to Gödel's physics and cosmology notebooks.

AE
Aidose Editorial
Nov 15, 2026 · read
Back to News