My writing colleagues will have special feelings for “basic writing” (i.e. a course for students not ready for first-year composition). The course is thus defined as remedial. To the extent that no one is really sure how to write in a college course in a world that includes generative AI, arguably all our writing courses are remedial. Here though we might think of Bolter and Grusin’s Remediation rather than grammar quizzes. With AI, what is being remediated is the compositional process itself.

With that in mind, going back to basics in a disciplinary sense. Can we still “name what we know”? I think we can. Writing is still

• Rhetorical
• Contextual/Situational
• Social/Cultural
• Ideological
• Material/Technological
• Practiced

I won’t attempt an all-inclusive list, but you get the idea. Writing remains a matter of concern, to use Latour’s phrase. It is not a settled matter.

In these terms, the perceived threat (and promise) of AI is that it will present writing as a settled matter. We can see this in a comically literal way when someone simply cuts and pastes AI output as writing. The AI has settled the task of writing for the user. But there are deeper and subtler effects on a cultural level of coming to view writing as a settled matter. We have heightened awareness of some of these dangers: hallucinations, bias, psychosis, and so on. And we have a growing awareness of other cultural worries such as the way AI output creates an implicit compositional convergence. A recent study of unscripted English-language podcasts identified convergence among the many different podcasters toward common AI phrases. These are rhetorical feedback loops and we have familiarity with them as these are the way that genres form in communities. However, when generative AI operates within the community, it will always dominate in raw output, and that output operates toward convergence.

It is worth noting that the AI is indifferent to our judgments about the settled quality of its output. Each new output operates as a settled matter. It never experiences (a lack of) confidence; it does not “revise” past output exactly as it has no experience of a past. It simply takes the old output and new input and operates to output settled matter.

However, if writing is not a settled matter, then we might reflect upon our tendency to treat it as such, to say to students “we know how you will need to write.” Because we don’t. We never really did, but now that fiction is played out.

What we do know about writing, as a practice, is that for the human writer writing occurs over time from a state of limited knowledge and in a distributed, cognitive environment that now includes AI. We also know that environment is all those bulleted things above, plus more.

While that writing environment has changed, the university curriculum generally has not. It still employs writing for the same purposes it did five years ago. Students remain positioned and oriented toward course writing in the same way, but it isn’t the same because the entire context has shifted. So obviously we will get different results.

The fundamental difference is the amount of human labor needed to produce text as a settled matter.

This is primarily a problem for academia’s process for evaluating and certifying students, which incorrectly assumes the amount of human labor needed to produce text as a settled matter. In the spirit of Douglas Adams, I would term this “somebody else’s problem.” I do not think it is the response-ability of rhetoric and composition to solve or salve the contradictions of neoliberal education.

I think we would all agree that while our discipline might add many more bullets and argue over those bullets, no one would be suggesting that what our field knows is that writing is a settled matter. Let me rephrase that. Writing isn’t a settled matter. It isn’t words set on a screen or printed on a page. So I wouldn’t accept response-ability for securing writing as a settled matter in order to prop up institutional fictions about learning.

So what is the difference between teaching writing, composition or rhetoric and teaching students to settle matters through media output. Well, for one thing, we wouldn’t ask student output to settle the matter of whether or not they have achieved learning objectives. We can verge toward the romantic and poetic and suggest that we might teach students to unsettle matters. Certainly some unsettlement is often part of learning and writing.

I would take a slightly different tack and suggest that we are discussing a different ethical matter, a different response-ability (in both the ways that Diane Davis uses that term through Levinas and Nancy and Haraway later does as an ethics of care that presumes an ontological foundation such as Davis’). Writing, for good or ill, is to be endured. It takes time. We can interact with AI as writers. In some ways we cannot avoid it in our media ecology. Can we learn to interact with AI so that it unsettles rather than settles things for us? Maybe. That’s the romantic version. Regardless, the ethical burden for writing falls upon us, not the AI. Allowing an AI to settle those ethical matters for us through its non-ethical output (note not “unethical”) collapses all our efforts.

If we are writers, then we have some capacity for action, and we must articulate how that capacity can be practiced to keep AI from settling matters for us, even as we acknowledge their participation in our media ecologies. Figuring that out for ourselves would be useful for discussing the matter with students. Regardless, none of us really know how we will need to write in five years. However, as faculty in the field, I’m sure that we have some ideas about how to deal with that uncertainty without pretending it doesn’t exist.