←back to thread

An LLM is a lossy encyclopedia

(simonwillison.net)
509 points tosh | 1 comments | 29 Aug 25 09:40 UTC | HN request time: 0s | source

(the referenced HN thread starts at https://news.ycombinator.com/item?id=45060519)
Show context
narrator ◴[02 Sep 25 15:32 UTC] No.45104411[source]▶
An LLM is a lossy Borges' Library of Babel

"Though the vast majority of the books in this universe are pure gibberish, the laws of probability dictate that the library also must contain, somewhere, every coherent book ever written, or that might ever be written, and every possible permutation or slightly erroneous version of every one of those books. " -https://en.wikipedia.org/wiki/The_Library_of_Babel

replies(1): >>45105158 #
RodgerTheGreat ◴[02 Sep 25 16:20 UTC] No.45105158[source]▶
It's a version of the library of babel filtered only to the books which plausibly consist of prose. The set is still incomprehensibly vast, and all the more treacherous for generally being "readable".
replies(1): >>45107936 #
1. dragonwriter ◴[02 Sep 25 19:32 UTC] No.45107936[source]▶
“Biased” more than “filtered”; LLMs, though its harder with the better ones, can, when prompted “right” (or “wrong”, depending on your perspective), produce output that is not plausible prose, though obviously they are crafted to respond to the expected style of prompting with things that are plausible prose.