by Zoe Tzanis
3 April 2026
This is not a praise of further AI use in academia. This is a warning to the AI-curious and to those already using AI in their everyday scholarly life.
I do not use AI tools in my writing. But recently I came into probing AI while revising—specifically when I received a particularly critical revise-and-resubmit from a particularly critical reviewer. After reading through several negative remarks and even some direct insults from one of my reviewers, I was at an absolute loss as to how I could possibly appease this very angry anonymous scholar.
I had outlasted my own knowledge reservoirs, as well as those of my advisor. What more could I do? As a last-ditch effort, I probed ChatGPT. I input the review comment and asked, “What can I do to please this reviewer?”
Chat answered immediately, spitting out a few lines I should add to the paragraph the reviewer was addressing, as well as a very, very generous, to the point of being obviously superficial and synthetic, response I could include in my response letter. I sighed. Not only did these additions fail to capture anything substantive, but they were also attached to several non-existent sources. Obviously, ChatGPT was no help.
So I tried again. With Claude. And what I got was miraculously useful. While Chat had hallucinated writing and posed it as something I should pass off as my own—which I would never do—Claude analyzed the reviewer’s language and generated a bullet-pointed list of key issues that needed to be addressed, along with a panoply of suggested revision approaches. It wrote nothing for me. Rather, it suggested: “You might want to add two sentences considering X theory and citing critical scholarship in this area,” and “This reviewer really detests this sentence, perhaps delete it or come up with a strong justification for keeping it.” It also offered an apology for the reviewer’s harsh words. Nice touch, Claude.
These suggestions gave me the ground from which to pose a response, but they did not shove any words in my mouth. Claude let me do the creative work and did the parsing of where to start. If you receive a review like this, maybe Claude could be of use to you. Maybe Claude could make it all worse. It’s up to you. You choose your scholar support team, and whether Claude or any AI belongs on it.