logoalt Hacker News

iLoveOncall • today at 3:56 PM • 0 replies • view on HN

Or maybe the tools are not good?

I can't write more than a few prompts without the LLM (any of the latest SOTA models) starts doing complete absolute nonsense.

I do agree that the ideas are not good to be clear, but the gap between the promises of what LLMs can do and what they can actually do is huge.

Anyway, it's funny to see the insane amount of astroturfing on HackerNews and how people try at all cost to make it seem like the technology isn't the problem but humans actually are. Literally every top-level comment is about that, or downvoted.