Post #4136420
2026-07-27 11:31 UTC
What does it even mean to "have just formalized a complicated paper in a language and not needed to know any of said language"?
Would anyone use DeepL and proudly proclaim to "have just written a complicated novel in Chinese without knowing any of it"?!
What's our epistemic understanding of "formalisation" here? Uninteresting busywork these programming people do?
Replies (2)
-
@ohad@mathstodon.xyz 2026-07-27 11:39
@gallais@mamot.fr that's my silver lining, that these developments will push all the busiwork and superficial tickboxing to such farcical extremes that we give up on pretending they are surrogates for quality or productivity.
-
@brillenfux@mastodon.social 2026-07-27 11:40
@gallais@mamot.fr I reject such use of LLMs exactly because of Fallibilism: Currently, nothing about- and of- their work is ever falsifiable. As such their output can (of course) inform intuitions, but not actually generate knowledge. And for informing intuitions, in most cases, they are WAY too expensive (on a global scale).