Gödel's lost workbook results surface as AI works through a logician's transcripts
Kurt Gödel, the Austrian logician behind the 1931 incompleteness theorems, left 16 notebooks from the early 1940s known as the workbooks. They run to 1,009 pages, much of it in Gabelsberger, a German shorthand already obsolete when he used it. Since 2017 Jan von Plato of the University of Helsinki has led a project on them, backed by a €2.5m European Research Council grant. He taught himself the script and says only six people in the world can read it. Progress was about one page a week.
That changed after he met computer scientists at an AI and theorem-proving conference in France in August. Von Plato gave Claude his own transcriptions, his six earlier books on Gödel and the secondary literature. Within a couple of days it produced a 650-page explanation of the workbooks. Asked to find theorems Gödel had proved, sketched or had ideas for, it found 663. Checked in the Coq proof assistant, about 400 were verified as stated, about 200 were repaired by tightening statements or adding assumptions, and about 50 were false.
The reconstruction suggests Gödel was almost at a proof of the independence of the axiom of choice. One definition was wrong, though he had the correct one two years earlier. Von Plato says the proof passed Coq and believes publishing it would have earned Gödel the 1940 Fields medal. Paul Cohen won that prize in 1966 for the independence result, after Gödel claimed to have proved it in the 1940s but could not rebuild a proof from his notes. Von Plato also says the workbooks anticipate dozens of results not known until the 1970s and 80s.
Why it matters: archival knowledge is often locked behind scarce human expertise. Here the human reading of the shorthand was still essential, but an AI turned transcribed pages into checkable mathematics far faster than one page a week.
The caveats are real. Von Plato says Claude's output must be checked page by page. Joel David Hamkins of Notre Dame has not seen the reconstruction and says how much is Claude and how much is Gödel is for historians to decide. He adds that resemblance to Cohen's proof does not mean Gödel had it, and that unshared work earns no credit. The Fields medal claim is von Plato's belief, and the figure of 14 logic themes not pursued since is Claude's own assessment.
Source: Guardian
Many Minded