Hacker News new | ask | show | jobs
by 1vuio0pswjnm7 267 days ago
Why can't "AI" be used to fix such errors

Unless this grammatical error is extremely common, LLMs should make this an easy one to detect (relatively low probability of next token)

Why not use "AI" to fix human error (like this one, presumably) instead of expecting people to fix "AI" output