In one of the papers I'm currently working on (not in Nature), I strongly suspect one of the reviewers used an LLM to write their review. Seems like they wrote 20ish comments suggesting minor grammar changes with only a handful of comments that actually address the content of the paper.
Answering comments/suggestions made by a chatbot rather than a subject expert is a waste of everyone's time and likely won't do what actual peer review is intended to do: set a quality standard and filter for the journal.