What even is writing?

6 min readDec 19, 2022

As a writing professor, I never looked for plagiarism. It looked for me. I would be reading along and suddenly a sentence or a paragraph would shift without explanation into a different register, as if a singer had jumped into an octave range they really couldn’t pull off. The effect was just as cringy and and dissonant. Who needed TurnItIn when my brain was already doing sophisticated pattern recognition?

Among the complex reactions I experienced in moments like these was bemusement. Regardless of why a student had plopped un-cited material into their paper, there was always a sense of how did they think I wouldn’t notice this? But actually, part of what it means to be a novice in any genre is not being able to distinguish between different patterns within a larger category (say, writing) or not knowing that there are patterns to begin with. To become an expert is to be able to make ever finer distinctions.

But at the point at which an AI can pattern match a student’s characteristic errors or write a paper on any subject to meet any assignment type, well, we have some bigger questions.

That point was apparently last week.

I’m saying this because when I asked ChatGPT to write me a slightly naughty story, it clutched its pearls and said that wouldn’t be appropriate. But when I asked it to write me an expository paper at a sophomore level about the environmental issues within the cruise industry, it said, no problem — here you go. It said the same thing when I asked it to write a post-structuralist analysis of a Shakespeare sonnet.

At that point, we have to ask not only how will we teach writing now, but also why?

At that point we have to ask what even is writing?

Right. So contemplating the entire future of the profession I trained for wasn’t really on my to-do list this week, but everyone else is doing it, so here we go.

What even is writing?

Writing, like most forms of communication, is all about creating (and also breaking) patterns. The patterns are what make it intelligible, useful, pleasurable. Gmail’s auto-fill reminds us over and over of how much of our written communication is utterly predictable. On the one hand, it’s a bit insulting; and on the other, you really were going to say that, so why not save the typing?

When we teach scholarly writing, so much of what we’re doing is teaching rhetorical patterns that many faculty (myself included) have found it useful to use They Say/I Say: The Moves that Matter in Academic Writing, a kind of pedagogical Mad Libs that helps students to learn to read for and to express the differences between their own ideas and those of a “source.” It helps students try on the patterns that make scholarly conversation possible. And it makes those conventions explicit rather than assuming that students will imbibe them through osmosis.

More importantly, we try to explain why it’s worth distinguishing between our ideas and those of others, why those quaint quotation marks and weird, exacting bibliographies (for the record, I’m totally fine outsourcing APA and MLA and IEEE formatting to the machine) are actually central to the whole project of knowledge generation. And we try to explain why it’s worth participating in those conversations even as a student — why it’s worth doing their own stunts, so to speak. Back in the day this mostly seemed to work with the occasional exception.

It’s hard enough to teach something that the culture itself doesn’t widely hold up as a common source of admiration or at least fame. I’ve often found myself thinking that the shame is not just that so many students leave college without learning to write well; it’s also that so few will ever need to.

But what about when the culture ceases to need writing skills in a more literal sense? Will we continue to teach writing like we do basic math — sort of ceremonially in case someone finds themselves on a desert island without a calculator…or an AI? Here are a few thoughts about how the teaching of college writing might change.

Composition Futures

Maybe we will look back with nostalgia at the era where we taught children and young adults to self-generate or auto-generate texts. A few purists will still write themselves, but it will be a niche activity like developing film or using a record player or typewriter.

Maybe what it means to become literate will no longer include the production of written syntax, but will instead mean becoming an excellent director of the AI. We will need a word for this.

In a sense, everyone will learn to write the genre that writing instructors and other faculty compose all the time: the writing assignment.

While the category of human writing still remains relevant, maybe students will be asked to learn generic conventions by becoming reliable raters of AI output.

Probably eventually the distinction will cease to be meaningful.

Maybe citation will come to mean learning to give the AI credit. Or the AI’s data set.

Maybe if students are taught to write, it will all be by hand in class. Maybe cursive will make a comeback.

Or maybe all student writing from kindergarten through the doctoral dissertation will be entered into and graded on a single program to which teachers are given successive, temporary keys. The model will be programmed to “age” along with the student’s growth as writer. Cheating will be impossible because the model will be trained on a combination of the student’s own writing and statistically forecasted improvement.

On the grading side, assuming that self-generated work continues to be a thing, what if a teaching AI could do a “first pass,” adding clear, customized instruction about common syntactic errors, leaving the instructor free to engage with the overall ideas and strength of the arguments (even though those are often hard to pull apart). God knows, I have composed some boilerplate text to use in my responses to student work.

Surely it could be useful in the writing of assignments. To kick things off, I asked ChatGPT to: