Scholarship in the Age of AI

Scholarship in the Age of AI

Tags: academia, musings

Published on Friday, May 29, 2026

" Previous post: Epistemic Humility in the Age of AI

While academic institutions are scrambling to pretend that they are at<br>least partially in control of the AI revolution, some more established<br>actors have already had to adopt new measures.<br>The venerable arXiv, for example, recently announced that they will ban authors for a year under certain circumstances:

If a submission contains incontrovertible evidence that the authors did not check the results of LLM generation, this means we can&rsquo;t trust anything in the paper.<br>The penalty is a 1-year ban from arXiv followed by the requirement that subsequent arXiv submissions must first be accepted at a reputable peer-reviewed venue.<br>Examples of incontrovertible evidence: hallucinated references, meta-comments from the LLM (&ldquo;here is a 200 word summary; would you like me to make any changes?&rdquo;; &ldquo;the data in this table is illustrative, fill it in with the real numbers from your experiments&rdquo;)

I believe that the policy is right in spirit but I am worried about<br>the way it will be imposed and whether there is due process for all.<br>That being said, in this post, I am more interested in a high-level<br>discussion on what it means to follow the norms of scholarship in the<br>age of AI.

In a nutshell, my position is this: You are ultimately responsible<br>for your work and cannot abdicate that responsibility to a tool.

The moment we start assigning personhood to an AI system,<br>it becomes a proper coauthor and collaborator, requiring direct credit<br>authorship. But until we are there,1 you cannot shirk your duties. It<br>is perfectly acceptable to use whatever tools you have at your disposal<br>to write your papers, but you are supposed to remain in charge. If you<br>use AI for literature search, for instance, you need to check the<br>results for (a) existence, (b) correctness, and (c) content.<br>You need to do this because scholarly work relies on citations the way good detective work relies on a chain of evidence.<br>You, as an author, have the duty to tell your readers about the relevant literature<br>landscape. It needs to be clear to what<br>extent you are extending the state of the art, relying on earlier<br>models or arguments, and so on. The existence of AI does not change this<br>fundamental prerequisite of scholarship.

Mistakes can happen and will happen,2 but not all mistakes are equal.<br>For example, referring to a non-existent work is highly problematic. It<br>erodes trust in your own work and, beyond that, the trust of the public<br>in science itself—at least to some degree. Referring to claims in<br>a work that are not part of that work is similarly problematic. You are<br>misleading readers while also misrepresenting the work of others. By<br>contrast, referring to a preprint instead of the published version of<br>a paper is, ultimately, harmless. The discourse around the<br>aforementioned arXiv policy misses this type of distinction, unfortunately.

Enough about bibliographies, though! Scholarship is much more than that,<br>but the same principle applies: You are ultimately<br>responsible for your work and cannot abdicate that responsibility to<br>a tool.

Here are some more concrete examples:

Using AI to (re)write your paper implies that you need to understand<br>editorial suggestions before accepting them.

Using AI to (re)write your code implies that you need to defend or<br>justify modeling choices.

Using AI to (re)write your proofs implies that you check their<br>correctness.

None of these examples ask whether it is a good idea to use AI for<br>these purposes. Since I lack concrete data, we are now entering deeply<br>speculative and personal territory. I am going to start<br>with a confession: I derive most of my enjoyment from mulling over things<br>and solving problems. Whether it is writing, reading, or coding, I just<br>love the process as such. Offloading certain tasks to AI robs me of<br>that joy; Terence Tao used<br>the following analogy in a recent Atlantic interview:

AI tools are like taking a helicopter to drop you off at the site. You<br>miss all the benefits of the journey itself. You just get right to the<br>destination, which actually was only just a part of the value of<br>solving these problems.

Since I do not know what, in the words of Marie Kondo, &ldquo;sparks joy&rdquo; for<br>you, I can only leave you with the generic piece of advice that you need<br>to decide when to take the helicopter and when to hike yourself.<br>However, when you do take the helicopter, make sure to (a) acknowledge<br>it and (b) check that it actually put you where you wanted and needed<br>to go in the first place.

We are all figuring things out in these times. Have courage to be truthful<br>to yourself, and the rest will follow.

Let&rsquo;s see how badly this statement is going to age! ↩︎

To err is human; to forgive divine; to never make a mistake at all<br>is reserved for reviewer 2. ↩︎