alt Hacker NewsNo doubt this is happening, but I wonder how many authors of papers "slated for desk rejection" 10 years ago could answer questions about their papers? We'd need that comparison to understand if this is a new problem or if AI is just a new source of content that the authors of poorly-written papers are using.
Evidence-based assertions are a good thing, but some things are so obvious that a "comparison" or "research" is not needed. This is already an obvious problem in so many places from high schoolers turning in assignments they don't understand, coders submitting code changes they don't understand, blog posts, and certainly to scientific papers.