At home, language is steam—

rice cooker clicking into rest,

my mother’s voice folding

another country into the

kitchen.

I answer her in English,

carry guilt like a second

lunchbox.

Robert P. Hubbard Assistant Professor of Poetry Michael Leong, with whom I took ENGL 291: Asian American Poetry, showed me a poem with this stanza. He had prompted ChatGPT to write an Asian American poem. “Here’s an Asian American poem — lyrical, reflective, and rooted in everyday inheritance,” ChatGPT proudly announced before producing the poem.

We had a good laugh about it. Any student in Asian American Poetry can tell you why the poem is offensive. “It’s a parody of what you think an Asian American poem is like,” Leong said.

Kenyon College is proud of its creative writing tradition. Has the emergence of AI systems such as ChatGPT changed this? I spoke with Leong, who teaches poetry, Assistant Professor of English Alyssa Quinn, who teaches fiction and Richard L. Thomas Professor of Creative Writing Ira Sukrungruang, who teaches creative nonfiction, to find out.

When I asked the professors whether they had ever suspected that a student used ChatGPT to generate their writing, I received a unanimous no. “Only once did the thought cross my mind, but ultimately, I didn’t think the story had been written by AI — only that it was the kind of story that could be written by AI, in the sense that it felt quite familiar and fairly tidy,” Quinn wrote in an email to the Collegian. 

In their class policies, these creative writing professors prohibit the use of AI, but they leave the door open for conceptual experimentation under the condition that the student is transparent about using AI-generated text.

The reason why Kenyon students in creative writing courses keep their distance from AI is not due to class policy, however. Leong and Sukrungruang credit the nature of the genres and students’ connection to writing. Poetry students often come to the class because they have written poetry in high school or because they are interested in songwriting, Leong said, “so I think there’s a real investment in actually doing the work and having a creative practice.”

In the case of creative nonfiction, that the genre is rooted in lived experience doesn’t lend itself to AI use, Sukrungruang remarked. He went on to emphasize the importance of embracing voice and its mercurial quality: “One of the things I spend a lot of time in my classes talking about is voice and what AI does is finding one singular voice. … In nonfiction, we talk about our fractured selves [and how] our voices shift and change within the page.” Students’ aversion to AI can also simply be emotive. “Most of my students find AI viscerally upsetting and wish to keep their distance,” Quinn wrote.

The professors each identified what large language models cannot replace in students’ creative writing practice. Leong talked of AI ‘s lack of cultural competency and how ChatGPT “boils down the poem to Chinese Americanness, or Vietnamese Americanness, or specific food items.” Sukrungruang echoed a similar sentiment before adding the loss of emotional nuance to the list: “It’s very black and white, and to me, literature exists in that gray space.” Quinn stressed the value of idiosyncrasies: “The most accessible AI programs — ChatGPT et al. — are bad at all the things I most value about writing — surprise, weirdness, nuance.”

The community aspect of writing at Kenyon is also highly incompatible with AI usage. I, like many students who have been in a creative writing workshop, understand the vulnerability of sharing your writing and the bond you form with the peers who are generous and constructive toward your works. “If everyone is doing their own work and you’re not doing it, it seems like a violation of the creative spirit of the community,” Leong commented. In creative writing class, students’ work is read multiple times by classmates and professors before maturing into its final form. Understandably, it’s difficult for a machine to intervene in this very human, communal process. “Because [students are] writing about their lives, I meet with them quite often. We have a lot of conversations,” Sukrungruang said.

The optimists among us might claim that they can immediately tell which writing is produced from human creativity and grit, and which is AI slop. Unfortunately, the truth is that the technology is constantly advancing toward producing sophisticated and sincere-sounding text, to the point that some people might prefer AI writing over human writing. A New Yorker writer recently had her friends and fans read four passages — half were her writing and half were AI-generated — and asked them to identify which is which. Her readers preferred the AI writing and thought it was hers. As confident as we would like to be that our humanity will shine through and distinguish itself, some models can mimic the brilliance and flaws of writers if fed enough input. 

The advancement of AI means that there’s a chance students’ usage goes undetected. Quinn and Sukrungruang are unfazed. When asked if there’s guidance they want to give to students who might feel tempted to use AI in any part of the process, they both responded that they are not looking for clean, perfect, error-free writing. “[Students] think there’s one particular kind of ‘good’ writing that they should mimic — but there isn’t,” Quinn wrote.

Sukrungruang expressed that as an educator, he cares about the “why:” “Is it about insecurity? Is it about competition, right? Is it about a sense that my art isn’t good enough? … What can I do to help you so that you don’t feel insecure?” 

Clearly, the creative writing instructors of Kenyon care about the integrity of the process, and they want to help students embrace the delight and messiness of it all. I walked away from the conversations with a renewed appreciation for the ungainly and deeply rewarding work of creating one’s own language instead of relying on a borrowed one, among a community of peers and instructors who similarly find joy in the writing process.