logoalt Hacker News

kolinko • yesterday at 1:08 PM • 0 replies • view on HN

Output of LLM can be infallible* even if LLMs themselves make mistakes. Ditto with humans.

As much as anything can be infallible.