ASK HN: SHARE YOUR AI PROMPT THAT STUMPS EVERY MODEL (HACKER NEWS THREAD)

Summary from the TLDR newsletter

Any prompt that has been fed into a public large language model has already probably been used for training, even if they haven't been disclosed publicly. This makes it difficult to create personal benchmarks for models - mainstream benchmarks already have a high risk of being gamed. This thread contains prompts that AI still can't answer, along with discussions about the inner workings of large language models. Examples include taking a common riddle and then changing one little detail, asking something about an obscure movie, and asking about things that don't exist.