To what extent should AI be used in scholarship? Which uses are okay? Which aren’t? Should AI ever be used? What impact will AI have on scholarship in the future?
I recently had a very insightful conversation about these issues with Nita Farahany (Duke Law) and Orin Kerr (Stanford Law).
Relevant to our discussion in the video, here are a few recent stories in the news.
Scholar or Machine?
One professor churned out more than 200 papers and 14 books in a year:
The University of Chicago statistician has authored or co-authored more than 200 papers and as many as 14 books released this year, unleashing speculation on the internet about whether he used artificial intelligence to inflate his production without disclosing the assistance. . . .
The online platform where Polson posted his papers said it removed 257 of his works over the past month after his surge in releases. Critics have not accused Polson of fraud, but have raised questions about whether Polson properly disclosed his use of AI and whether it’s even possible to produce that many high-quality papers in so short a span.
A Glut of Papers
The number of papers submitted to academic journals has increased dramatically in the age of AI:
Will journals be able to cope with the tsunami of article submissions? It will make the jobs of journal editors much harder, and it will make it harder for scholars to get published.
How Frequently and Extensively Are Academics Using AI in Scholarship?
An study by Oxford University Press, “Disclosing AI Use in Academic Research: A Global Study” (Sept. 2026) has many interesting findings about the use of AI. It appears that the vast majority of scholars are using AI to some degree.
How Well Can AI Write and Entire Article?
Professor William Baude (Chicago Law) tried an experiment and used a chatbot to write a law review article. He prompted Claude to write a law review article expanding upon a post he wrote and to write the article in his voice. He discusses the results in a post:
I don’t think it is as good as the piece I would have written and I hope you don’t either. I did not write the piece, and there are plenty of places where I think I would have written it better. And I am not sure I agree with all of the piece, since — again — I didn’t write it.
Still, I thought the results were quite good. Better, frankly, than I expected. It even has a Baude-like joke in the last sentence of the * footnote. (I also enjoyed “it is the Court’s doctrine, not mine, that presently cannot tell a crocodile from a ballroom” and “IV.B. Other People’s Babies”).
Although the piece was entirely AI generated, he notes: “When I ran the article through Pangram, it came back as only 48% AI-generated.”
Ironic Stories of AI Use
There’s an ironic story about a scholar’s use of AI in a book that has been generating a lot of news. The irony is that the book is called The Future of Truth: How AI Reshapes Reality (2026). Although he disclosed the use of AI, several quotes in the book were hallucinations:
The author of a nonfiction book about the effects of artificial intelligence on truth acknowledged on Monday that he had included numerous made-up or misattributed quotes concocted by A.I.
The author, Steven Rosenbaum, whose book “The Future of Truth” was released this month to great fanfare, incorporated more than a half-dozen misattributed or fake quotes in sections of the book reviewed by The New York Times.
The Times asked Mr. Rosenbaum about the quotes on Sunday and Monday. On Monday night, Mr. Rosenbaum acknowledged in a statement that the book had “a handful of improperly attributed or synthetic quotes” and said that he had started his own investigation.
He said that the inclusion of the incorrect quotes was an accident and that he had “no intention of fabricating any viewpoints” while writing the book.
A big risk of using AI for anything is the risk of hallucinations. It can often be more tedious and time-consuming to doublecheck everything AI does than simply to avoid using AI.
In another ironic case, a student newspaper accused the school’s provost of using AI in scholarship to point out hypocrisy of penalizing students for AI use:
College professors and administrators have been busting students for cheating with artificial intelligence for several years. Now, the professors and administrators are getting in trouble, too.
Dartmouth College last week announced it was investigating its provost, Santiago Schnell, after the college’s student-run newspaper found that some of his published writing appeared written by A.I.
This case raises another danger of AI. Even when it might be used appropriately, others may find that use inappropriate or immoral. Scholars must weigh the risk that however they might come down on whether certain uses of AI are okay, there will be some who will view such uses as bad, and there’s a risk that scholars might find themselves unwittingly castigated in a viral news story.
Unstable Norms
Should scholars be credited for articles using AI? It strikes me as fine to use AI to help with research, hone ideas, format citations, and do proofreading. It strikes me as wrong to use AI to draft entire articles or even significant portions of them. But what about drafting a small section which the scholar edits extensively? It depends. I feel differently depending upon whether the section were just background information or an overview of cases versus an interpretation or argument. It’s hard to know where to draw the line.
I view scholarship as a conversation among scholars, and I read and write scholarship to be part of this conversation. When AI plays a minor supporting role, such as proofreading or even helping to hone prose, it doesn’t bother me. It’s AI acting as a copyeditor, and I’m still in conversation with the author. But when AI is the engine of the ideas and arguments, then AI is becoming the primary author or co-author, and I am not reading scholarship to be in conversation with a bot. If I want to talk to a bot, I can do so directly myself.
Other readers may appreciate only the articles, and the authors behind them really don’t matter. As long as the articles are good, it doesn’t matter how they were produced or who wrote them. But I’m still in love with the romantic vision of the author. When I read a text, I see myself as engaging with a person through an artistic medium. As corny as it sounds, I believe scholarship can and should be a form of art.
This doesn’t mean that AI can’t produce scholarship or art. It can. And I wouldn’t reject scholarship or art on this basis. I think of it like playing chess with a human or machine. I can have a good chess game with each, and I certainly want to know whether I’m playing a human or machine. I don’t see the machine replacing the human even though it might be better. The same, I hope, will be true for art and scholarship. There is something different about engaging with another human mind, and it is something that shouldn’t be lost.
I’m writing fiction these days, and the norms in the world of fiction appear to have solidified rather quickly: Zero use of AI is tolerated. Here’s guidance from the Author’s Guild which is quite restrictive but still about as generous towards AI as it gets in the literary world. Most journals and book publishers view AI as a plague and want not even a whiff of it. In one recent instance, a book publisher revoked a multi-million dollar book contract on suspicions of AI use.
For scholarship, the norms are far less figured out—they’re as unstable as uranium. Until they settle, writer beware!
* * * *
Daniel J. Solove is the Bernard Professor of Intellectual Property and Technology Law at the George Washington University Law School. He is the founder of TeachPrivacy, a company that provides workforce privacy, cybersecurity security, and AI training to companies and organizations around the world. He is the author of 10+ books and 100+ articles.
You can follow his events, writings, training, cartoons, and resources by subscribing to his free weekly newsletter.
Subscribe to Solove’s Free Substack
A supplement to Solove’s regular newsletter with more in-depth discussions
