July 29th, 2026
There’s a new scholarly paper available, published July 22, 2026, titled “Generative AI floods and dilutes the market for books.” It might have remained obscure, as academic papers often do, except that Alexandra Alter made it the focus of her July 28 New York Times article, “How A.I. Books Sneak Their Way Into Stores.”
Alter is the author of the well-known exposé of the Shy Girl case, published March 19, 2026, titled “A.I. Is Writing Fiction. Publishers Are Unprepared.” In that article she referenced Tuhin Chakrabarty, a professor of computer science at Stony Brook University, who is the main author of this new paper. In her article she reported that he had “used Pangram [AI-detection software] to check more than 14,000 self-published novels on Amazon for A.I. writing. The program found that nearly 20 percent of the novels had been substantially written by A.I.” I wrote to Chakrabarty after reading Alter’s piece, asking for more information about his study, and he replied on LinkedIn “it’s not public yet.” Now it is.
The paper is thorough and well-documented. It’s thoughtful and considered. I think that it’s also way off the mark, and, in its own way, misleading. As the paper seeks, in part, to address a major issue regarding the potential damages in the current litigation surrounding how AI companies ingested vast quantities of copyrighted text, it bears further scrutiny.
What Does the Paper Say?
As noted, this is an academic paper. In its PDF format it’s 31 pages long. Exported to Microsoft Word, it totals nearly 17,500 words (including all footnotes, sources and captions). It contains some intimidating mathematical formulas, like this
and this
No conclusions are stated, but in the abstract of the paper the authors state “Generative AI can… reshape a creative market through scale rather than quality. Our results bear directly on the market-effect question at the center of the fair use defense to copyright infringement.”
OK: So What Does the Paper Say?
Reduced ad absurdum, this is how I read the paper: Let’s say a bookstore carries 10 different titles by self-published authors. Those will sell a certain number of copies, perhaps a few, perhaps a bunch. Now, what if the bookstore also starts carrying two or three books generated, at least in part, using AI? Some people will buy those books, perhaps because they don’t realize that they were produced using AI, which might affect their quality, or perhaps because, regardless, they seem worth reading. As a result, those 10 titles by self-published authors will sell fewer copies.
Bookselling is essentially a zero-sum game. Every book that sells has sold at the expense of another book that didn’t sell, but might have sold if it faced no competition. The publishing industry in the U.S. hasn’t grown significantly in decades. Same number of books, same number of dollars. My sale gained is your sale lost. (A big footnote here about just how wrong that statement is in the statistical details. But the point still stands.)
And so, in that sense, Chakrabarty’s paper is accurate. New books on Amazon, whether created with AI or not, cannibalize the sale of existing books on Amazon. Oh, but my god, there are already so many books on Amazon, so very many. Do AI-generated books truly upset the apple cart?
“A.I.” is used as a presumed pejorative throughout this article. There’s a built-in assumption that books created with AI are inferior to those written without AI. But what if, as a thought exercise, we turn that on its head, and build in a bias that books generated with the help of AI tools are consistently of a higher quality, and offer greater reader satisfaction, than those created without. Now the article would be arguing that good books are crowding out bad books, and trying to demonstrate why that’s an undesirable outcome, as it negatively impacts the careers of bad writers.
The authors of the article might say, no, we’re not taking a stance one way or another — we’re only trying to demonstrate that books written with the help of AI are cannibalizing the sale of books written without — there’s no implied judgement. Well then, why would you even bother with the pursuit? Ah, yes, because it’s a major topic in the current AI copyright litigation. Yes, that explains it. This is an issue before the courts, not before the public. You are weighing in on a legal issue. What is it, exactly, that you hope to gain?
50 Shades of Grey: The Use of AI in the Authoring of Books
Ah, yes. The hornet’s nest. We could say that there are as many opinions as there are authors and readers. But that’s not really true. Distilled to their essence there are only two opinions: the use of AI in authoring is a bad thing, or the use of AI in authoring is a bad thing.
OK, there are shades of grey. There’s the most-often-heard, “well, it’s ok to use AI for spell-checking and basic grammar checking. But anything more than that…” But the general consensus amongst the literati is that AI most absolutely must never be used to generate text for publication.