THE IMPOSSIBLE PROMPT THAT'S EASY FOR HUMANS

Summary from the TLDR AI newsletter

This post presents a rare prompt with a simple, objective answer that is trivially verifiable, in which LLMs fail 100% of the time, but most humans can produce a correct answer.