'Positive review only': Researchers hide AI prompts in papers

They are professional researchers and doing the reviews is part of their professional obligation to their research community. If people are using LLMs to do reviews fast-and-shitty, they are shirking their responsibility to their community. If they use the tools to do reviews fast-and-well, they’ve satisfied the requirement.

I don’t get it, really. You can just say no if you don’t want to do a review. Why do a bad job of it?

I dunno. There generally isn’t super high security around preprint papers (lots of people just toss their own up on arxiv, after all). But, yeah, it is something that you’ve been asked to look after for somebody, which is quite important to them, so it should probably be taken pretty seriously…

I dunno. The extent to which, and the timelines for, the big proprietary LLMs to feed their prompts back into the training set, are hard to know. So, hard to guess whether this is a serious vector for leaks (and in the absence of evidence it is best to be prudent with this sort of thing and not do it). Actually, I wonder if there’s an opening for a journal to provide a review-helper LLM assistant. That way the journal could mark their LLM content however they want, and everything can be clearly spelled out in the terms and conditions.

But, I also expect that eventually every prompt is going to be a candidate for being added into the training set, for some future version of the model (when using a hosted, proprietary model that just sends your prompts off to some company’s servers, that is).

That's why I mentioned it. Worrying about training on the submitted paper is not the first thing I'd think of either.

When I've reviewed papers recently (cancer biology), this was the main concern from the journal. Or at least, this was my impression of the journal's concern. I'm sure they want to avoid exclusively AI processed reviews. In fact, that may be the real concern, but it might be easier to get compliance if you pitch this as the reason. Also, authors can get skittish when it comes to new technology that not everyone understands or uses. Having a blanket ban on LLMs could make it more likley to get submissions.

Every researcher needs to have their work independently evaluated by peer review or some other mechanism.

So those who "cheat" on doing their part during peer review by using an AI agent devalue the community as a whole. They expect that others will properly evaluate their work, but do not return the favor.

"Improve the model for everyone - Allow your content to be used to train our models, which makes ChatGPT better for you and everyone who uses it."

It's this option that gives people pause.

But, I think it is worth noting that the task is to make sure the paper gets a thorough review. If somebody works out a way to do good-quality reviews with the assistance of AI based tools (without other harms, like the potential leaking that was mentioned in the other branch), that’s fine, it isn’t swindling or defrauding the community to use computer-aided writing tools. Neither if they are classical computer tools like spell checkers, nor if they are novel ones like LLMs. So, I don’t think we should put a lot of effort into catching people who make their lives easier by using spell checkers or by using LLMs.

As long as they do it correctly!

If someone is just, like, working chatGPT up to automatically review papers, or using Grok to automatically review grants with minimal human intervention, that’d obviously be a totally nuts thing to do. But who would do such a thing, right?

Edit, consider the following hypothetical:

A couple of biologists travel to a remote location and discover a frog with an unusual method of attracting prey. This frog secretes its own blood onto leaves, and then captures the flies that land on the blood.

This is quite plausible from a perspective of the many, many, ways evolution drives predator-prey relations, but (to my knowledge) has not been shown before.

The biologists may have extensive documentation of this observation, but there is simply no way that an LLM would be able to evaluate this documentation.

I wouldn't specifically use either of those words because they both in my mind imply a fairly concrete victim, where here the victim is more nebulous. The journal is unlikely to be directly paying you for the review, so you aren't exactly "defrauding" them. You are likely being indirectly paid by being employed as a professor (or similar) by an institution that expects you to do things like review journal articles... which is likely the source of the motivation for being dishonest. But I don't have to specify motivation for doing the bad thing to say "that's a bad thing". "Cheat" manages to convey that it's a bad thing without being overly specific about the motivation.

I don't have a problem with a journal accepting AI assisted reviews, but when you submit a review to the journal you are submitting that you've reviewed it as per your agreement with the journal. When that agreement says "don't use AI", and you did use AI, you cheated.

That's a self-contradicting statement. It's like saying mass warrantless surveillance is ethical if they do it constitutionally.