←back to thread

Hallucinations in code are the least dangerous form of LLM mistakes

(simonwillison.net)
371 points ulrischa | 2 comments | 02 Mar 25 19:15 UTC | HN request time: 0.492s | source
1. cryptoegorophy ◴[03 Mar 25 00:51 UTC] No.43237104[source]▶
Just ask another LLM to proof read?
replies(1): >>43239693 #
2. namaria ◴[03 Mar 25 08:43 UTC] No.43239693[source]▶
Do you realize that giving LLMs 'instructions' is merely trying to blindly twist knobs by random amounts?