Charlotte Aten

It seems that @zbMATH has introduced #LLM output into its system. According to https://zbmath.org/static/newsletter/zbNEWS-21.pdf the organisation has used a chatbot to convert OCR output from its first 529 volumes into #TexLaTeX. Disappointing, but perhaps I shouldn't be surprised. It's published by #Springer, after all.

#academia #research #math #mathematics #journal #publishing #AI #enshittification #zbMATH

5:46pmWeb
Show threadDima Pasechnik 🇺🇦 🇳🇱13h ago
@caten that is how the OCR systems had been working since their appearance over 35 years ago. They have to be trained on samples of text ever since. @zbMATH
Show threadCharlotte Aten13h ago
@dimpase I know how OCR works. They explicitly say that they passed the result through an LLM to produce LaTeX.
Show threadDima Pasechnik 🇺🇦 🇳🇱13h ago
@caten and? It's a special purpose LLM. You can get an open-source LLM (or write your own if you wish) and use it to improve OCR.
Show threadCharlotte Aten13h ago
@dimpase I think I had to tell you I wasn't down to debate on here one other time. Bye.