Tim’s got some really clever tests that he’s throwing at the latest batch of LLMs. His conclusion is that while they’re good at language and words, they fall flat in some pretty major ways elsewhere.