在生成式 AI 时代，我们究竟看重写作教学的什么价值？

Over the last nearly four years, I’ve done well over a hundred keynotes, workshops, public talks, podcasts and print interviews on the problem of trying to teach writing in a world in which generative AI is present.

I go to universities, colleges, high schools, public forums of varying stripes. I’ve zoomed with school boards, book clubs and classrooms of pre-service English teachers. I talk to students, faculty, teachers, parents, administrators—not enough administrators, frankly—anyone who reaches out and says they’d like to have a conversation.

I have spoken to exactly zero elected officials. Curious, that.

Most Popular

• DOD Ends Partnership Between Brown, Naval War College
• Anti-DEI Trump Official to Lead DOJ Civil Rights Office
• Bill Supporting College Students With Disabilities Advances

As challenging as things still appear to be, I can report that by and large significant progress has been made in terms of figuring out how to work this problem of teaching writing in a world in which generative AI is present. The schools and higher ed institutions that are doing the best on this problem have embraced a collaborative approach where all parties—students, instructors and administrators—are working the problem together.

At the heart, everyone’s interests are aligned. Students want to learn, teachers want their students to learn and administrators are pleased if students are learning. Parents, too, where they’re part of the equation.

The biggest challenge, then, is all the things that stand between students and learning. When it comes to high schools, I am increasingly concerned that structural factors around so-called accountability constrain teachers from adapting to the challenge. The number of times I have modeled an approach that I believe works to focus students on learning and been told something like, “This sounds great, and I’d love to do it, but …” is concerning.

But … as I said, progress. I think about this infamous article from May 2025 in New York magazine that declared “Everyone Is Cheating Their Way Through College.” “Cheating” was the primary concern or frame of reference at that time. From the beginning of this AI era, I thought this was an ultimately unproductive frame and said so in a 90-minute conversation with the author of that article, prior to the article’s publication.

I was not quoted in the article, probably because every answer I gave to his questions started with “It’s complicated.”

It is complicated, by the way.

From “everyone” is cheating, we fast forward to a couple of weeks ago when Princeton University announced that they were providing free ChatGPT subscriptions to all and a flood of students took to the student paper and said, essentially, “Please don’t do this; we want to learn!”

What I’ve come to realize more and more is that our discussions about how to adjust to a world with AI are not so much a matter of policy—to purchase ChatGPT or not to purchase ChatGPT—or even pedagogy—i.e., where should we incorporate AI?—but values.

If we don’t understand and share our assumptions about the underlying values, we cannot even begin to have a productive conversation about how to put those values into practice.

Editor's Picks

• Manhattan Institute Targets Mellon for Funding Left-Leaning Higher Ed Groups
• 4 in 10 Students Flag Digital Wagering Risks 
• Women Leave Academia Feeling Devalued, Study Shows

As a couple of root examples, I fundamentally believe that the core mission for a higher education is to focus on student learning.^[1] Others will tell you the focus must be on job preparation.

I believe grades should reflect how much students learn in a course. Others believe grades are best used to rank students against a standard, or each other.

Who’s right? It’s complicated, and hugely dependent on course, context and overall goals.

I want my future doctors rated against a standard, a high standard, before they are allowed to become doctors. While they’re learning to be doctors, however, I think I want them focused on how to learn, because I expect my doctor to be a true, lifelong learner.

The people who invite me to talk or workshop or podcast or whatever tend to do so because they already share my values, often in an unstated way. They read this column or my books and they think to themselves, That guy knows what he’s talking about. It is great to be affirmed, but it has its limits when it comes to broader progress.

Sometimes when I commune with these people, we sit around and wonder why everyone can’t see these things the way we do. It’s so obvious!

But it is obviously not obvious, because the people I work with seem to be something of a minority in the larger discourse around what we should be doing about the presence of AI and teaching writing. That the vast majority of these people are teachers of writing and/or writing studies scholars—disciplinary experts, if you will—suggests that we are not doing a great job of articulating these values.

Mea culpa—though, honestly, More Than Words: How to Think About Writing in the Age of AI is a pretty solid effort on this front if I do say so myself—but that just brings me back to the problem of the people who read you already agree with you.

So, I’m just going to lay things out in a kind of bullet-point format. This is the framework of values I attach to learning how to write, along with a little bit of why these are my values.

• Writing is a great way to, for lack of a better phrase, figure things out. Writing is an experience of thinking that requires the writer to grapple with an idea, and in that grappling to change and hone and ultimately possess that idea in a deep way. This is what makes writing worth doing, even as LLMs seem to (emphasis on seem to) give us answers to any questions we might have.
• The big-picture goal of learning to write in a school context is to develop a robust and adaptable writing “practice” consisting of the skills, knowledge, attitudes and habits of mind of writers. A good practice allows the writer both to continue to grow their capabilities inside a familiar genre and to transfer their existing capabilities to unfamiliar or novel writing occasions.^[2]
• Writing is a kind of problem-solving where a communication is needed for a particular occasion, targeted toward a specific audience. This is what is known as the rhetorical situation: message, audience, purpose. All writing happens within a rhetorical situation. (There are other considerations in terms of targeting a rhetorical situation, e.g., genre, but they can all be folded under those three broad categories.)
• Learning to write is learning to work inside authentic rhetorical situations where the writer practices the fundamental skill of writers, making choices that satisfy the needs of the rhetorical situation.
• For this reason, I believe all writing done by students should be framed inside an authentic rhetorical situation where the writing product is assessed according to its effectiveness at solving the problem of communication inside that rhetorical situation. (To see what this looks like across a variety of academic and scholarly artifacts, see my book of curriculum, The Writer’s Practice.)
• Writing, to a teacher as a mere “grader,” is not an authentic rhetorical situation and hinders the potential progress of student development.
• Rubrics are not an aid to helping students develop problem-solving skills inside authentic rhetorical situations because we don’t write to rubrics, we write to audiences. When we write, we should not be thinking of abstract principles like sound organization or clear expression, but instead be focused on the needs of the audience as they encounter our communication. (I should probably do a whole column on the problem of rubrics, but for now, I’ll refer people to Rethinking Rubrics in Writing Assessment by Maja Wilson and note that this book was published in 2006.)
• Providing students structural templates to fill in the content of a piece of writing (e.g., the five-paragraph essay) may help them secure a better grade measured against end-product criteria, but it denies them the opportunity to practice making choices about sequence and structure that are vital to development.

(Metacommentary note: Right now, I’m really struggling with what order to put these bullet points in, because there’s a few different paths I could take.)

• Valuing reflection and understanding of the process is central to student development in a writing course.
• Relying on the quality of end product as the sole criteria for assessment ignores the most important part of the equation when it comes to improving as a writer: the process. (See above.)
• The feedback and conversations between student and instructor should primarily focus on surfacing the choices the student has made inside the rhetorical situation and building student understanding of the effectiveness (or not) of these choices. The written artifact is necessary evidence to engage in these conversations, but it is most important as a way to discuss the writer’s choices, not to judge the outcomes.
• Because writing is ultimately the act of making choices, the use of AI in ways that prevent the student from engaging with the struggle of that choice-making hinders their development as writers.
• Because writing is meant for other unique intelligences, we should not allow AI to assess student writing.
• Speaking very personally here, I have no use for feedback from generative AI models because they do not read, think or feel and cannot communicate with intention. The remixing of syntax generated after being prompted by my writing is just not anything that is helpful to me in my writing process.
• Finally, back to the beginning, writing is an embodied experience, and when it comes to learning to write, I want students to have the deepest, most embodied experiences, which is why I believe the exchange must be between humans without AI involvement.

If we’re going to discuss and disagree how we might want to deploy AI in writing, we have to start at the level of these values, otherwise we’re talking past each other.

I’ll be honest, I don’t anticipate moving off these values, given that they’re something I’ve lived and built over 30-plus years, but I will also admit that in a world where there is pressure to deliver outcomes, my values may be out of step.

I also accept that there are people who believe that AI has obviated the need to maintain this experience of human writing. I think they are off their rockers, but I accept that their view is sincerely held.

I also know that if we are in a realm where there is a general belief that learning to write does still matter, we should not invite the people who think it doesn’t matter to the conversation. They have their own thing going on.

How do I end this? I don’t know. In frustration, I guess that we haven’t made even more progress, but in hope that the opportunity for more progress remains.

^[1] In my book Sustainable. Resilient. Free.: The Future of Public Higher Education, I put it more expansively: “I believe postsecondary education should aim to develop the intellectual, social, and economic potential of students.”

^[2] In More Than Words I write about how doing a poetry explication on a Gerard Manley Hopkins sonnet in a form and theory of poetry class in grad school ultimately prepared me to write a long, quantitative questionnaire when I was working for a market research firm. This is when I discovered I had a writing practice. I only wish someone else had told me this before I approached 30 years old.

注意转载授权 · 阅读原文