"Peer Reviewed" by ChatGPT: Motive, Opportunity, and Means

As you might notice, I co-author a lot of papers. Thus I see many peer reviews. I am also a heavy ChatGPT user for lots of things. One thing I have noticed in the past year is the number of reviews “written” by ChatGPT.

Some may be the reviewer doing the review and then running it through ChatGPT for polish. I have some sympathy for this, many reviewers in my field do not find English to be their native language, so this saves time on writing something more than bullet points (but do we need more than bullet point?).

On others, however, I think the “reviewer” just took the PDF of the paper, submitted it to ChatGPT, gave it a query like “review the attached paper”, and cut and paste the results into a review form without making any changes. Editors seem not notice or care — or maybe they themselves are these “reviewers” — Maybe the editors themselves do this three times, with three different queries to generate three different takes, and we have a complete “peer review” done in a timely way, to which the author must respond to somehow? It sure is less work for the unpaid and unvalued reviewers. So we have motive (save time), opportunity (do this anonymous Peer Review, it’s hard to prove), and means (ChatGPT). And this just leads to more busywork on the part of authors who need publications for promotion and status to satisfy the bean

counters.

So we have now become slaves to the AI, responding to its probabilistically random mutterings in the name of scientific process and scientific progress.

Dall-E Drawing of the above: Here’s the illustrated scene capturing the humorous yet reflective view of AI's role in academic peer review. It showcases the blend of relief, frustration, and detachment among reviewers, editors, and authors as AI-generated responses shape the process.

FIN.