Yet, I was left wondering about a potential issue in the underlying premise and the graph presented: it could well also be because the AI field is booming, not only because people are using AI for "paper engineering". A similar point applies to the arXiv's recent policy change; it could well also be that OA is gaining more and more traction. And it could further be that pre-prints are gaining more and more traction precisely because the peer review system is more or less broken in the big picture of things.