In “What even is an author?,” Katherine Mangan and Shea Vance grapple with a troubling question that the increasing use of AI in scholarly writing is raising at universities: At what point does authorship even matter now in academia?

A recent article in Philosophy & Public Affairs by University of Hong Kong philosophy prof Simon Goldstein illustrates the conundrum:

Goldstein disclosed his use of AI in the article’s “acknowledgements” section and in a separate five-page report detailing how Claude contributed each step of the way. “The paper was an experiment in writing with AI,” he wrote “ — in how far Claude could go as, in effect, the Ph.D. student, with the author acting as something like a very generous Ph.D. supervisor.”

After Goldstein told Claude the “basic idea,” it “wrote almost all of the prose in the paper,” he wrote. Goldstein told The Chronicle that he would have listed Claude as the first author and himself as the second had the journal allowed that. He did almost none of the actual writing, but said his intellectual contributions to the paper — editing, correcting, and guiding Claude — met the threshold of authorship. September 11, 2026

Goldstein defends his approach: “The chief purpose of academic journals, he said, is disseminating new knowledge, and if AI can do that, it’s unreasonable to prohibit its use just to protect the secondary function of credentialing scholars.” But he also told The Chronicle that Claude has not yet come up with a good original idea for a paper, so it is not clear what role new knowledge would play.

Many academic publishers are trying to set limits

But to what extent are they failing to grasp the nature of the problem: If academic papers in a discipline can be mostly AI written, why is there a need to credential scholars anyway? Perhaps the discipline itself is obsolete.

At Big IFF True, Daniel Muñoz predicted a number of changes that the AI revolution entails, including,

1. Quantity of research is probably going to matter less for job prospects and promotion.

2. Quality of research will probably matter even more in the near future, since the best humans are still hard to beat.

3. “Impact” of research will also matter more, which means more pressure to game whatever system we use to measure impact.

4. The field is going to scramble to find alternative ways to assess philosophical ability beyond the traditional research paper. Expect departments to put more weight on interviews, rec letters, job talks, and Q&As.

“Is Claude an agent of the death of philosophy itself?,” September 14, 2026

Time will tell. But the key problem is, if only truly original work matters, how large will a given scholarly discipline be down the road?

So what did Jason Arday do that was so wrong?

Recall the recent Jason Arday tragedy/scandal at Cambridge. The principal accusation against Arday was plagiarism, copying someone else’s work and passing it off as his own.

Back in the days when a scholar was assumed to have labored far into the night over many years on his own work, plagiarism would indeed be unethical. But what if, down the road, a scholar has in fact gotten a chatbot to compose most of the work? Arday’s only offense would be that he had not bothered to fire up the chatbot to do the same thing himself. That would be a lapse, sure — but how would it meet the standard of intellectual theft that the traditional understanding of plagiarism covers?

Reaching a bit deeper into the questions…

If chatbots start to invade more areas of writing for publication, how important will copyright continue to be? Many will argue that the books that chatbots produce would be of inferior quality and lack original ideas. That’s probably true. But, as Orwell noted nearly a century ago, much of the market is not looking for original ideas, just the ones readers like and are used to. Copyright will be harder to defend if only a small and unrepresentative group of authors with original ideas still need it.

Schools are currently battling to reduce the use of AI because they feel the need to teach students to think and write for themselves. But some will ask, do students still need to learn that stuff? Can’t they just learn to do what scholars are starting to do?

Some will go so far as to argue that AI will be a great leveler in education. Even students who do not grasp a topic can sound like they know it. Of course it’s true that the student has not really learned the work. But social promotion is an old story in education. And how important is learning anyway if thinking is outsourced to AI? How many people will continue to think it is a problem if students come to resemble the Eloi in The Time Machine (1895) — they know nothing and are happy?

If the academic world slowly adapts to AI-written scholarship as the way of the future, it’s hard to see how all of these trends will not follow. And we will end up asking, with good reason, who needs scholars?