Authors of seemingly low-quality TMLR submissions reportedly did not understand their own papers

A post describes Nihar Shah’s interviews with authors of seemingly low-quality TMLR submissions, saying he spent 20–25 hours over two weeks on the effort. It claims those authors had “no idea” what was going on in their own papers. A comment quoting that account argues that heavily AI-generated papers are less likely to have competent authors behind them. The commenter also says they have little patience for anything Pangram identifies as mostly AI-generated.