I was on the fence about reading John Warner’s new book, More Than Words: How to Think about Writing in an Age of AI. It wasn’t that I didn’t respect Warner’s work. No, I had been following his work for years, especially his Just Visiting blog on Inside Higher Ed where he writes about teaching. His was an essential voice for me after ChatGPT launched in late 2022. He had already argued in his book Why They Can’t Write: Killing the Five-Paragraph Essay and Other Necessities that writing instruction in secondary and post-secondary schools was failing. In his December 2022 newsletter, he noted that if ChatGPT was good at writing the kinds of essays we ask students to write, then maybe those essays weren’t worth writing to begin with:

I quoted that in many of my talks and workshops on teaching and AI throughout 2023!

No, I was on the fence about reading More Than Words because a book about the implications on teaching and learning of a new technology that is published just over two years after the widespread availability of said technology is quite likely to be premature. I had read more than one book on teaching and AI that wasn’t fully baked. And I have more books to read than time to read them, so I try to be selective about which books I add to my shelf of reading opportunity. What convinced me to pick up Warner’s book was a comment of his on Marc Watkins’ newsletter this past January. Here’s what Warner wrote:

That’s a challenge I can get behind! In fact, I feel that much of my work in higher ed over the last two years has been about helping faculty and other instructors figure out (individually and collectively) what about generative AI should be resisted, what aspects of teaching and learning should be valued regardless of AI, and how we might find useful roles for AI in our work as educators. This is, for example, why I’ve produced a dozen episodes of my Intentional Teaching podcast on the topic of AI and teaching. Educators across higher ed are struggling to figure these things out, and it is helpful in that process to hear from others who have been doing just this kind of exploring.

I finished reading More Than Words a couple of weeks ago, and I’ve been struggling with how to write about it. Warner is a fantastic writer, and he shares many stories from his own writing life and from his classrooms about the value of writing as a way to think and to feel. He argues compellingly that we shouldn’t outsource our writing to some AI automation tool. “To abandon all writing to generative AI is to abandon thinking itself,” Warner writes on page 68. And on page 78 he writes, “What happens when we divorce the necessity of feeling from the act of writing by outsourcing it to something like ChatGPT, which cannot think or feel?” I largely agree with Warner on these points; as James Lang writes in an endorsement, we should resist allowing “automation to supplant the work that makes us human.”

My problem is that there is a fair patch of middle ground between “abandon[ing] all writing to generative AI” and resisting the use of AI altogether. And it’s in that middle ground where I see thoughtful and caring educators spending their time exploring the intersection of AI and teaching.

For example, consider the ChatGPT-as-peer-review assignment that Pary Fassihi shared during her interview on Intentional Teaching. In the assignment, she asked students to request feedback on their essay drafts from ChatGPT using very specific prompts tied to the criteria Fassihi had set up for the essay assignment. Here’s one: “Evaluate the evidence used to support the main argument. Is the evidence relevant, sufficient, and effectively integrated into the argument?” Here’s another: “How well does the paper analyze the implications of digital technology on academic integrity and authorship? What insights or unique perspectives does the paper offer?”

There’s some evidence (Steiss et al., 2024) that AI chatbots can give feedback on par with human feedback when directed via criterion-based prompts like this. Warner cites this study, then dismisses it because he didn’t find the essay assignments in the study compelling. “The assignments used in these studies are lousy, canned prompts that have been answered by students a zillion times with no room for individual engagement or knowledge building” (page 240). The assignments are available in the paper’s appendix, so readers can judge for themselves. I find them a little cookbook, in that they are fairly prescriptive about what writing moves students should be making. But in a way, so is Fassihi’s assignment, given the clear criteria she laid out for her students through the suggested prompts.

If we can believe for the moment that the assignment at hand is a reasonable one, Warner is still resistant to any kind of AI feedback on student work. He writes the following on page 260:

However, consider the instructions Fassihi gives her students on this assignment. She specifically asks them not to make all the changes that ChatGPT recommends:

Fassihi goes on to ask students to revise their essays and also to write a short reflection about how they used or didn’t use the ChatGPT-generated feedback. There’s no imperative here to “please the algorithm.” On the contrary, Fassihi is making an effort to explore with her students when and how AI might provide useful feedback while also taking a critical look at AI and its limitations.

That’s the kind of explorative work that I thought Warner was calling for in More Than Words, but there are very few examples of it in the book. The one memorable example is Warner’s tentative embrace of using AI as a reading assistant of sorts. Page 120: “ChatGPT and other equally powerful models are especially good processors of text, and sometimes, having a tool that can process a lot of text very quickly and mostly accurately can be a useful substitute for some of the ‘reading’ we are often required to do.” Warner goes on to share a story about using ChatGPT to generate an outline of a book he had read several years prior, then using that outline to conduct a targeted re-reading of the book, saving Warner the time of a full skimming of the book.

That is a great example of using AI as an assistant instead of an automation or outsourcing technology. I talked with an education faculty member last fall who had a student doing an independent study who used AI in a similar way. The student was reading journal articles she found very hard to read, so she asked Google NotebookLM to produce one of its chatty-podcast-style audio summaries of one of the articles. That summary provided the student the big ideas and structure of the paper, which made it easier to dig into reading the full paper after listening to the summary. In this case, as in Warner’s example, the AI didn’t replace the reading of the text. Instead, it played a helpful role in that reading.

I can think of other examples from the guests I’ve had on my podcast. Heidi Nobles talked about ways to use generative AI as an editor when writing, getting an approximation of the kind of feedback a human editor or a writing group might provide. Ryan Wetzel shared experiments in having AI serve as a catalyst for students working through a design thinking process, helping students with divergent thinking or rapid prototyping. And way back in November 2022, before ChatGPT was released to world, I talked with Robert Cummings about AI writing, and he pointed me to the band Yacht, who used an AI tool to generate lyrics and music based on their own catalog of songs as well as on songs from their favorite artists. They took the pieces of the AI output they liked and wove those into new songs for a new album. The AI became (to borrow Wetzel’s term) a catalyst for the group’s creativity.

Warner is right to push back against what he calls AI “true believers,” the Sam Altmans of the world that make hyperbolic claims about the power and impact of AI. But I learned a long time ago not to pay much attention to the wild claims of technology companies. I would much rather explore the technology itself and see where it might be useful. I was hoping for more of this kind of exploration in More Than Words, in part because examples of explorations like the ones above were out in the world at the time of Warner’s writing. More Than Words has a lot to say about the power and value of writing, but by limiting its treatment of AI to AI-as-automation and not fully considering ways AI might serve as a catalyst for thinking, the book doesn’t provide the roadmap to the future of AI in writing (and learning) that I hoped it would.

Thanks for reading!

If you found this newsletter useful, please forward it to a colleague who might like it! That's one of the best ways you can support the work I'm doing here at Intentional Teaching.

Or consider supporting Intentional Teaching through Patreon. For just $3 US per month, you can help defray production costs for the podcast and newsletter and you get access to Patreon-only interviews and bonus clips.