And what's so tiring is that with humans, there's often (not always) a coherent meaning there to be untangled. With the LLMs, probably not, because if there was, it would have been in the prompt and the rambling wouldn't be as bad.