Do you suppose there's any connection between how LLMs write and how humans write?