It’s certainly already made it worse. And I have been knocking back a lot of papers that at first look sound amazing but on deeper reading say nothing of value. The common trend now is to make a paper sound like you have done the experiment when actually it was just a thought experiment or simulation. I have seen theory papers with diagrams you would expect from an experimental students PhD thesis.

It’s having a terrible impact on the review process. I have been getting clearly LLM responses from reviewers for manuscripts, emdashes and flowery language all over the place but saying nothing of value. Which sucks because decent reviewers are often a big help when progressing research.

But the bigger issue is that in general I think the review system is overwhelmed. I recently got a single line reviewer response for a manuscript submitted to Optics express, a journal I would have considered above such issues in the past. The quality of review is in freefall right now.

I have talked to colleges around the country that feel the same way. I don’t think the existing system will continue in the old way much longer. At this point, youa re almost better off putting a groundbreaking discovery straight on the arXiv and just skipping the peer review process. It is basically just a waste of time now, and only still exists as a gatekeeping step into prestigious journals. I also look at younger researches with high h indexes suspiciously. How much time did you spend gaming the paper system as opposed to actually doing useful research that takes time but generates less papers?

“Look, how much GenML fucked that field into the ground, essentially killing it!”
“WHAT?? SOMEONE FINALLY FOUND A USE FOR IT??! Hans, MAKE A TOOL FOR THAT SPECIFICALLY!”