@UrsSchreiber I'm so sorry. It's definitely negligent. You might write a polite complaint to the publisher and/or the scientific board to inform them of the editorial negligence but of course that won't change anything about your submission. Maybe(?) they'll say something to the editors.
We've never had a non-human which was able to generate coherent text, and even proofs. I don't think it's so clear that they're "just" tools, or that we deserve credit for what they do when we prompt them. I think it raises a number of important philosophical questions that
@prof_g And maybe the answer is to try to make your analogy better - to add different kinds of credit for different roles in the process: "Director", "Guy who solved problems in detail", "Paper writer", "Editor", "Robot assistant", etc.
@prof_g Except that it's an important philosophical point for the profession, as it affects how we will assign credit going forward - and, as a result, jobs, grants, and everything else.
@littmath @d_m_d_m_d_d @CoFore856 Really? Most academics outside the US don't have an .edu email... And plenty of non-researchers inside the US do. If that's really true, it's absurd.
@prof_g Have you ever watched the credits for a movie, though? I'll accept your analogy when we have coauthor categories equivalent to "screenwriter", "actor", "director", "music composer", "stuntmen", etc
@prof_g I'm surprised when people say - and even deeply believe - that they wrote something when all the actual words were produced by an LLM. I strikes me as an odd and largely incorrect extension of one's sense of self.
@prof_g But they are both brainstorming ideas, proofs, and even actually writing parts of the paper, if I understand correctly, right? If a person contributed what they do to your papers, wouldn't that person be a coauthor?
@prof_g@lu_sichu If you've ever had to read someone's research code in order to understand a model they incompletely explained in a paper, I'm not sure you'd be looking forward to this future of refereeing. Still, I hope it'll be an improvement.
@prof_g@lu_sichu Lean (or other proof-checkers) will hopefully speed up reviewing, but they won't be enough to make it instant. One will still need to check that the human-readable results and proofs in the paper are in fact what was programmed into Lean.