My guess is that the fraction of humans who understand enough about LLMs to ask those questions overlaps pretty heavily with the fraction of humans who know that LLMs are almost universally terrible at accurately answering questions about themselves

→ View original post on X — @simonw