Recently I raised the subject of ChatGPT with a group of Ph.D. students, on the cusp of faculty careers. Which writing task would they like to hand over to a bot? And which one would they rather not farm out to AI? Their general rule: Use ChatGPT for routine tasks, and save personal writing for yourself. Yet their views began to diverge as soon as we started talking specifics.
What about cover letters for job applications? Those are supposed to be personal, yet the students split on whether to consign that task to AI or go it alone. The same was true of social-media posts. Of scholarly writing, most said they would never use AI — for, as one put it, “anything that requires argumentation and critical thinking” — but at least one student was even content dispatching that task to ChatGPT.
In academe, our debate has focused on the implications of generative AI for our teaching: How will we suss out cheating? Are there creative ways of incorporating AI tools into our courses? But we’ve paid far less attention to whether, and how, our own writing will be altered in the shadow of AI.
ChatGPT gets most of the media hype, but a host of other AI tools have preceded and followed it: Think of Microsoft Editor, Grammarly, Sudowrite, Wordtune, and now Google’s Bard, OpenAI’s GPT-4, and Anthropic’s Claude 2. As the models become increasingly sophisticated, we will at some point need to concede that neither human readers nor AI-detection tools can reliably discern if a text was written (or edited) by a person or a computer. AI as writer and editor is upon us.
We hire professors for disciplinary expertise, not writing prowess. Unless they’re in journalism, creative writing, or literature, we don’t expect colleagues to bring home a Pulitzer or a National Book Award. In fact, there’s no guarantee that even those teaching English composition are good writers. I’ll not forget the writing instructor who admitted that, while she could explain rules of grammar and style to students, she considered herself a weak writer.
Use of AI in faculty work raises troubling questions, but the ethical lines guiding academic writing and editing have been blurring for years. Some academics pay a professional editor to tidy up their drafts of articles or books. Software programs can reliably polish anyone’s syntax, punctuation, and style. AI also offers useful solutions. For example, for people whose second language is English, AI tools provide a handy, even essential, assist. Their potential extends beyond faculty work to administrative labor: I’ve seen colleagues decline to become department chair because of uncertainty over their English-writing skills. AI tools could ease that dilemma considerably.
Underlying the concerns about how AI will alter faculty work is a much larger and longstanding problem: We undervalue good writing in higher education. And that has consequences for both our teaching and our own writing.
There’s always been a sizable disconnect between our institutional mantras about the importance of students’ writing skills and our expectations for the quality and clarity of faculty writing. The now-ubiquitous English-composition requirement began entering American colleges in the mid-1880s. Almost 140 years later, at least a semester or two of college writing remains a rite of passage. Yet it’s an ill-kept secret that students’ writing skills commonly decline as soon as they finish those required courses.
Why? Because most faculty members (outside of English departments) grade a paper for its content and arguments, and don’t take it as their responsibility to critique its organization, grammar, and style. Students quickly pick up on the fact that good writing doesn’t matter, and respond in kind by paying little attention to it when they submit a paper. When a faculty member violates that tacit pact — by poring over students’ papers for organization, grammar, and style problems — that can backfire. I’ve had multiple students inform me that it was my job to comment on the content, not their writing.
With the emergence of AI, the pigheaded among us have a new challenge: A student’s written work can now look both coherent and squeaky clean, and we have no idea how much of that is due to ChatGPT and other such tools. And so the perfect storm:
- Most faculty members don’t relish investing time in grading or policing students’ writing.
- Students resent or ignore extensive comments on their writing.
- And AI can produce or improve upon what’s being turned in.
The result: a missed opportunity for all of us. At its best, writing isn’t just a concoction of words but a place for contemplation and wrestling with ideas. We talk incessantly about education fostering critical thinking, but far less about how the writing process advances that goal.
What is the solution? And can faculty members be part of it?
I’m not advocating that we all aspire for a Pulitzer or transform our courses into ersatz English-composition classes. What I am advocating is a much-needed overhaul of how we value writing in our scholarship and in our teaching. That need predates the AI revolution but becomes yet more vital when the likes of ChatGPT can seduce writers into relinquishing control over their words. And it will require buy-in from institutional catalysts such as department chairs and deans, and from administrative units such as writing programs and teaching centers.
What follows are some small changes that, together, could engineer a change in campus culture to recognize the value of writing as a tool for thinking:
How to value faculty writing. We’ve given years of lip service to the importance of nurturing undergraduate writing skills, yet the powers that be have rarely provided writing assistance to faculty members or graduate students. This doesn’t have to involve a huge investment. How about establishing a writing-support program for academics, including doctoral students, and offering resources for native and nonnative English speakers alike? How about giving release time from teaching to a skilled writing instructor who could work directly — and anonymously, if so requested — with faculty members on their writing.
Another simple step: Organize interdisciplinary writing partnerships for professors. Building on the model of informal writing groups (typically used for collaborative book projects), colleagues in different fields could be paired to read one another’s drafts of journal articles or book chapters, with the focus on writing clarity, not research content. Since we volunteer our time as reviewers in our academic disciplines, surely we can offer a hand to colleagues across the quad, and benefit from their insights on writing in return.
And of course, faculty scholarship in the age of AI needs guidelines. With the arrival of ChatGPT, institutions are hustling to formulate policies on student use of AI in homework, research, and writing. Typically, professional journals serve as watchdogs for faculty submissions, but some recommendations on AI use in faculty publication would be welcome additions to campus statements.
How to pay sustained attention to teaching writing skills. Step No. 1 is for faculty members — regardless of discipline — to talk with students about the role that writing plays in the thinking-and-learning process. You don’t have to be a literary type to have encountered Flannery O’Connor’s declaration that “I write because I don’t know what I think until I read what I say.” Spell out on your syllabus that both what students write, and how they write, matters in your course, and make it so. Emphasize that writing is a skill that needs to be practiced, and that you are interested in what they — and not an AI editor or text generator — have to say.
Part of the conversation should include that writing is a process. It’s not one and done. Writing entails rewriting. Engage with students about their likes and dislikes — and their fears — about writing, including honest discussions about how much they have come to rely on digital genies. Online editing tools like Microsoft Word or Grammarly have been mopping up our spelling and grammar for years. Now a host of AI tools recast entire sentences and, of course, write the whole shebang for us. The danger is for students to devalue their own efforts. As the Modern Language Association and the Conference on College Composition and Communication warn in their working paper on writing and AI, “students may not see writing … as valuable since machines can mimic [human] skills.” We need to bolster students’ confidence in expressing themselves in writing.
Which leads to Step No. 2: Again, regardless of discipline, have students do some writing in class — by hand. Peter Elbow, a professor emeritus of English at the University of Massachusetts at Amherst and a writing teacher par excellence, helped pioneer a practice called freewriting. The idea is to just sit and write, for a stretch of time, whatever comes to mind. Don’t worry about sentence mechanics, only about getting your thoughts out.
The technique is common in many college writing classes. But what if we introduced freewriting into the first few minutes of other types of courses? At the start of a semester, give all students bluebooks (they’re still cheap), and ask them to write what’s on their mind about the course — the homework, the readings, the coming tests. (As role models, faculty members need to join in the freewrite exercise as well.) Collect the bluebooks two or three times a semester to check that students have been writing, but don’t grade them.
Is that a sound use of precious class time? Yes, on several counts. Curricular models like the flipped classroom have taught us that enhancing student engagement in the time we have together is more productive than packing in yet more information. In their freewrites, students rev up their mental engines for the day’s session. Plus, the bluebooks give faculty members a periodic window into how students are responding to the course, potentially far more helpful than end-of-term course evaluations.
Another golden benefit of freewriting is nudging students to write by hand. In research I’ve done with young adults on what they like (and dislike) about handwriting versus using a digital keyboard, respondents spoke time and again about connections between handwriting and thinking. Among their comments: “I can see what I’m thinking” and “It leaves tracks in my mind.”
How administrators can help. In my former life as an associate dean, department chair, and later head of my university’s teaching center, I became acutely aware of gaps between what administrators knew and what information had made its way to the faculty. For tackling writing and AI issues, we need better communication and collaboration.
Start by letting faculty members in on the fact that undergraduate writing skills commonly decline once students finish their college writing requirements. (I learned that only because my comp and rhetoric colleagues clued me in.) Have departmental discussions on how to give students a reason to care about their writing in subsequent courses.
Tap your college writing program to share rubrics on how to assess student writing. Such guidelines are wins both for instructors (they are easy to use and don’t require buckets full of time to apply when commenting on student compositions) and for students (who will continue hearing that how they write matters). This model has worked at my own university, through a partnership between our college writing program and our teaching center.
Then there’s the importance of getting faculty members up to speed on what kinds of AI editing and author tools are out there, and how students might be using them. A succinct, balanced précis — coupled with campus policies and recommendations — will help individual instructors decide what’s suitable for their own classrooms.
Finally comes the trickiest piece of advice to administrators: Support a just mechanism for giving faculty members the time needed to confer with students about their ideas before setting fingers to keyboard, and then for teachers to read more than one iteration of a written assignment. Admittedly, doing so in a class of 50 students is a huge challenge. Even in classes of 12, it’s still a major time commitment. But if we are ever to know who did the writing (human or AI), these are necessary steps. More important, they tell students that we believe in writing as a process, not just a product. A learning experience, not just a grade.
For more than a century, American higher education has struggled to nurture student writing skills. Valiant efforts through English-composition courses and writing-across-the-curriculum programs have made their dents, but still left swaths of graduates skeptical that writing matters. The arrival of generative AI hampers faculty prospects for convincing students (and perhaps ourselves) that writing is a cognitive tool, a means of discovery, a uniquely precious form of human expression.
Because of AI’s incredible skills as author and editor, we have reached a tipping point. As academics, we can position ourselves (with some administrative support) to refocus the conversation about writing. Mustering the will to do so is on us.
