I just heard the best description of generative LLM AI yet: it’s programmed to create what an answer to your prompt would look like. Not an actual answer, so if the two overlap at all, it’s just luck. If that also involves things that are true, even luckier.