Hacker News new | ask | show | jobs
by bandrami 144 days ago
I still think that's missing the point of LLMs; they're sources of plausible text continuations. Their strength is using them in places where the actual semantics of their output isn't important.