Did a Chatbot Write a Prize-Winning Story? Does It Matter?
In May, the Commonwealth Foundation announced five regional winners for its Short Story Prize, recognizing unpublished fiction. One winner, Trinidadian writer Jamir Nazir, faced accusations of AI-assisted cheating after social media users flagged his story's synthetic patterns: repetitive sentence structures, strained metaphors, and passages that bordered on nonsensical.
The Commonwealth Foundation said contestants had confirmed they did not use AI, and authors made this attestation twice. The foundation's director-general later acknowledged the moment had come to examine whether their process was "robust enough."
When Nazir's story, "The Serpent in the Grove," appeared in Granta magazine, a Wharton professor ran it through an AI-detection platform. The tool flagged 100 percent of the text as likely AI-generated. Two other winning entries faced similar scrutiny.
Nazir said his writing process relies on speech-to-text dictation on an Android phone due to chronic health conditions that make sustained typing impossible. Granta's publisher said they asked Claude about the story's provenance but couldn't determine whether the judges had awarded a prize to AI plagiarism.
What the story reveals about AI detection-and human writing
Using chatbots to detect whether prose was written by chatbots carries its own problems. A Stanford study found that AI-detection algorithms show bias against non-native English speakers.
Still, AI-generated writing often exhibits recognizable patterns. Nazir's story contains anaphora-words repeating at the start of successive clauses-and epistrophe, words repeating at the end. Lines like "No fan, no bulb, no hum" and "Water took her and would not return her" fit the pattern.
The story also relies heavily on zeugma, where a verb takes one literal and one figurative object. Air becomes "sweet with cane and forgetting." A well's mouth is "boarded with ply and chance." These devices exploit learned associations between repetition and heightened meaning, but in AI text they disconnect from actual content.
Negative parallelism-the "not x but y" construction-appears throughout: laughter "cuts a hush, not cures it," and "bush took him in-not like a mother, like a judge." This structure is ubiquitous in AI prose.
When bad writing looks like AI-and what that means
"The Serpent in the Grove" contains glaring failures. A woman has "the kind of walking that made benches become men." Yet most problems are subtler: maudlin imagery, portentous descriptions of mundane observations, clichéd post-colonial melodrama.
The story treats inanimate objects as conscious agents. "The grove remembered." "Water is jealous." "Wood complained." A woman is described as "big in the way of women who never apologize to furniture."
Human characters lack specificity. They embody generic tropes: a pathetic husband, a maternal village woman, a subjugated bride. "Nineteen and brown like dust after rain, she turned roti dough with a rhythm that came not from joy but from endurance."
Compare this to V.S. Naipaul's description of a house in "A House for Mr. Biswas," set in the same location. Naipaul captures specific details-red curtains reflecting on polished floors-and infuses the scene with a character's actual envy and bewilderment. An AI struggles to generate comparable mastery of physical space or genuine human emotion.
The real scandal: what the prize reveals about literary culture
Many commenters focused less on whether Nazir used AI and more on why judges approved the story at all. The literary establishment, they argued, has been deteriorating long before AI entered the picture.
One judge praised the story's "precise yet richly evocative" language and "vivid, lush imagery with remarkable economy"-language that itself sounds formulaic, like a creative-writing seminar checklist.
Others saw the prize as evidence of condescension toward the Global South. Prize committees have been accused of rewarding aestheticized representations of poverty and mysticism in post-colonial fiction.
A writer created a parody of Nazir's tale, mocking how prize judges reward certain stylistic formulas. The satire worked because similar writing-romanticizing rural struggle and hardship-does garner praise from literary gatekeepers.
What AI exposure tells us about writing itself
The prospect of AI generating prize-winning fiction unsettles us because it reveals something about how we evaluate writing. We worry about writers who produce work for external rewards rather than internal conviction-hacks or courtiers seeking money and admiration.
External motivation produces writing that flatters readers' preconceptions. It shows romantic versions of the reader: now you're a sad child in a neglected home; now you're a soulful type discovering unexpected tenderness. It doesn't test readers by showing them who the author actually is.
If "The Serpent in the Grove" sounds like bad writing, that's because the AI was trained on bad writing. Asked to produce post-colonial fiction worthy of a literary prize, it delivered an off-key imitation of the genre's worst tropes.
As one critic observed, the scandal isn't that AI disrupts literary taste-it's that AI exposes how codified the formulas already are. The existing expectations for "authentic" post-colonial prose are so formulaic that a language model can reproduce them convincingly. AI "does not disrupt literary taste so much as expose its furniture."
Where AI poses the greatest threat
The risk of AI-generated fiction is sharpest in genres relying on stock characters and familiar scenes: romance and fantasy novels. Romance authors have been caught forgetting to remove AI prompts from published work. A writing coach teaches aspiring romance authors how to use AI, with over 1,600 clients including authors who publicly oppose computer-generated fiction but privately sign up for her classes.
When a horror novel was accused of using AI in its creative process, readers largely didn't care. The book maintained strong ratings on Goodreads even after the controversy broke.
The argument for AI in fiction amounts to: people have always written formulaic books, and if readers engage with it without complaint, how bad can it be? But readers deserve better. An AI cannot mean what it says. A human writer can agree with AI-generated text, aspire to it, or hide behind it-but she cannot mean it.
Writing devoid of inner purpose cannot rival work "ripped out of an author's chest with a claw grapple." Any serious reader knows the difference.
Your membership also unlocks:
