Being an academic doesn’t mean that everyone thinks you’re smart, I like to say; it means that a few smart people think you’re wrong. Addressing yourself to those people is the primary purpose of academic writing. That is, the purpose of academic writing is not to impress your peers (and certainly not your teachers) but to open your thinking to their criticism. This is why it is so important to do your own writing. By articulating your thoughts in prose — by “prosing your world” — you prepare yourself to learn what you are wrong about.

This isn’t as dramatic as it sounds. We expect academics not only to be knowledgeable but to be corrigible. They maintain what is known in an explicitly “logical” form, so that the consequences of changing our minds become conspicuous. It’s not that we’re perfectly deductive in our thinking, but if we’re wrong about one thing, we’re probably wrong about others, and “knowledgeable” people are able to more quickly and easily discover those other errors and carry out the necessary repairs. That is, a knowledge-able person isn’t someone who holds (and professes) a bunch of true beliefs. It is someone who knows how to be wrong — how to admit it and move forward.

What consequences does this have for our style of writing? Well, first of all, we must be clear about what we actually think. The writing must clearly assert facts, i.e., claim that things stand in one way and not in another. It must also clearly state our reasons for believing those facts. These reasons are individual points on which the reader may disagree with the writer, potential errors that, by being made explicit, are now easy (for another knowledgeable person) to identify. In other words, academic writing must be clear about what we think the facts are and why we think they’re that way.

That is one of the reasons we should carefully cite our sources. Much of what we believe we believe because we have read (and believed) someone else. Our reader, however, may have read the same source and had their doubts. By declaring whose authority we’re invoking, we open ourselves to criticism of that authority. More subtly, but just as importantly, we open our reading of the source to such criticism. That is, our reader may have read and believed the same source but understood it differently. We write in order to expose our interpretations of each other’s work to criticism too.

Like I say, I want this to be an argument for doing your own writing. It is possible that a language model can predict exactly what you would say about certain subject, and therefore, in principle, expose the ideas those words would express to the criticism of your peers. But that criticism won’t touch you in the same if the ideas were never form, in those exact words, in your own mind. You will be unprepared to listen to the critique. If you find the criticism persuasive you will now have to reconstruct the argument the machine generated for you in your own mind before you can adjust your thinking.

Now, it’s possible that Calvino’s fantasy (horror?) of a “writing machine surrounded by the hidden ghosts of the individual and of his society” can serve this same purpose. But we must keep in mind that this would mean generating a series of variations and testing each one by reading it until we achieve a “particular effect of one of the permutations on a man endowed with a consciousness and an unconscious, that is, an empirical and historical man,” i.e., the academic whose writing we’re proposing to replace. This effort of internalizing artificially generated prose is surely more demanding (and more risky) than articulating what you really think yourself.

There’s been a lot of concern about the effect of generative AI on higher education lately.* I share most of these concerns, though I don’t like to frame the problem in terms of student “cheating”. To be sure, many students are using AI in ways that allow them to earn better grades than they deserve — their exam performance overestimates their actual competence — and in that sense they are cheating, often quite deliberately (in knowing violation of rules). But, as I’ve long argued, it’s almost impossible to design an exam or assignment that doesn’t encourage students to pretend to be a little more knowledgeable, a little more intelligent, a little more competent, than they actually are. They may write as though they have read an entire book that they may only briefly have skimmed. They may write as though they’ve read a play when they’ve only read the Coles Notes or the Wikipedia article. Ideally, this would be detectable in the superficiality of their treatment, but surely you’ll admit that we often let them get away with it.

Indeed, as scholars, we often let each other get away with superficial readings of the literature — with prose like a context widow, i.e., prose that might as well have been generated by a large language model. Let’s admit that many of us think of scholarly discourse as an “attention space” and that it is our job mainly to make a “contribution” to it. We’re always looking for a “gap” and fit words to fill it with. Research is less the art of satisfying our curiosity than the business of coming up with something to say. George Orwell, of course, suggested quite a different image: our prose should be like a window pane through which we provide a clear view on what we are thinking. Writing is not merely the act of stringing together a “predictable” sequence of words in order to bring about a series of conventional effects in the mind of the reader. If, as Wittgenstein suggested, “we make ourselves pictures of the facts,” our writing presents those pictures to our peers for their careful consideration and critique. As I had occasion to explain a few years ago, the conceit of the scholar is not that their writing represents the facts but that we have a number of thoughts about those facts.

Norman Mailer once described “ego” as the “state of our psyche that gives us the authority to tell us we are sure of ourselves when we are not.” It is relatively easy to be brave in writing (which no doubt explains the famously toxic environment of social media. Would that conscience more often made cowards of us, let’s say!) Today, armed with a chatbot, students have it very easy indeed. They can write confidently of many things they know nothing of, indeed, things they have no real opinions about. So can we. But, now more than ever, we must write with the intention of exposing our thoughts to the criticism of our peers, and we must encourage our students to do the same. Only by telling each other what we actually think, not what it would be (statistically) “normal” to say, can we learn whether what we believe is true or false. We must be sure of ourselves long enough to find out whether our confidence is warranted.

I have decided to return to blogging because I need a space to work out what I think the future of academic writing can and should be. I agree with those who think that much of the crisis of AI is a result of our prose being out of shape, weakened by too many decades of “publish or perish” and “advanced placement,” unprepared to compete with the machines. But I don’t think all is lost yet. Mailer’s remark about the “ego” was inspired by Mohammad Ali’s “Fight of the Century” with Joe Frazier. I feel like it’s time to step back into the ring.

______________
*See Megan Fritts’ “A Matter of Words” (The Point Magazine); Zwi Mowshowitz’s “Cheaters Gonna Cheat Cheat Cheat Cheat Cheat” (Don’t Worry About the Vase); D. Graham Burnett “Will the Humanities Survive Artificial Intelligence?” (The New Yorker); James Marsh’s “Everyone Is Cheating Their Way Through College” (New York Magazine); and Phil Christman’s “Of Course Some Will Cheat” (Slate).