[gwern]

I understood his joke just fine. I just disagree, because I think he missed that it was AI-written in a way which is actually already implementing his suggestion of 'lowering' the original high-level human input into a more rigid, formalized schematic, easily-parsed language, which I call "AInglish", and which we increasingly are all communicating in, as exemplified by OP, due to the fact that many people find it superior and more enjoyable and more understandable, as is not too much of a surprise given the goals of the chatbot training process and cognitive offloading/deskilling. (As the joke goes, the LLM will expand out my short prompt for an email or blog post, and their LLM will summarize it for them...)

[fsckboy]

his joke does not rely on the truth value of the mistake you think he made, and it is funny based on his expressed awareness that any articles potentially could have been written by AI, so his error could simply have been a misplaced word typo/grammar error rather than any cognitive omission on his part.