Jamir Nazir, a 62-year-old Trinidadian writer, has won the overall Commonwealth Prize for his short story "The Serpent in the Grove" - less than a year after the same piece was accused of being AI-generated. The Commonwealth Foundation announced Tuesday that Nazir's story was selected from among regional winners after an investigation satisfied judges that he wrote it himself.
How the investigation unfolded
The controversy began in May, when Nazir was named a regional winner and his story appeared in the literary magazine Granta. Readers and critics questioned the prose, and the AI-detection tool Pangram scored the text as 100 percent artificial. Phrases such as "the kind of walking that made benches become men" circulated online, drawing mockery. Other winners soon faced similar suspicion. The Commonwealth Foundation initially defended the authors, then launched a deeper review.
That review did not use AI detectors. Razmi Farook, the foundation's director general, said the investigation relied on "detailed discussions" with each writer about their creative process and on "working drafts, time-stamped documents and notes" that showed how the stories developed. Nazir said he provided earlier drafts, character sketches, and other materials voluntarily to clear his name.
Writing with a phone and speech-to-text
In an interview after his win, Nazir described a writing routine shaped by his health. He has diabetes and neuropathy that makes sitting at a desk and typing painful. Instead, he writes on a couch using the speech-to-text function of his Android phone's Google keyboard. "That's what's actually producing words, and then I edit them and so on," he said. The small screen - about three and a half inches of visible text - forced him to hold lines in his mind and polish them repeatedly.
Nazir also stressed that the story drew directly from his life. "Why the hell - sorry for the expression - would I need AI? It was absolutely not needed in this story, because I lived this," he said. The bread shop in the story, he noted, is a real place in the village where his wife grew up. Writers who rely on speech-to-text tools often find that the dictation process shapes their rhythm and word choice in ways that can look unusual on the page.
Poetry, AI, and literary influence
Nazir pointed to poets such as Pablo Neruda, Gabriela Mistral, and Derek Walcott as influences. He described his admiration for Walcott's disregard of traditional Western sentence structure, a trait some readers had flagged as a sign of machine writing. He said he has written a collection of short stories in a Walcott-like style. When asked to name a favorite Walcott work, he could not recall one, attributing the lapse to brain fog from chemotherapy.
He suggested that AI-detection tools may produce false positives because they have been trained on the same literary influences he draws from - and, he added, on his own poems posted in Facebook groups. "AI must have been fed all their work," he said. "They have all of that as a reflection of my style as well, right?" He questioned whether an AI system that knows a writer's patterns might then flag that writer's original text as machine-made. Experts note that detectors like Pangram look for statistical regularities common in AI output, but Nazir's point highlights how fast these tools can misread highly stylized prose.
AI as the next typewriter
Nazir predicted that AI will eventually find acceptance in literary circles, much like the typewriter and the word processor before it. "When the typewriter was first invented, writers kicked hell and said, The thing is writing. You're supposed to use a quill or your fountain pen," he said. "Modern-day word processing can do almost everything. Where's the uproar? It settled down. Now I think the same thing will happen with AI." Still, he cautioned writers to stay away from AI tools in competitions for at least two or three years while the debate rages. He repeatedly insisted he did not use AI for his prize-winning story.
Why this matters for writers
The episode underscores three practical realities for anyone producing literary work today. First, false accusations based on AI-detection software can escalate quickly, and the best defense is a well-documented creative process - dated drafts, notes, and revision histories. Second, unconventional prose alone does not signal machine authorship; health constraints, dictation habits, and deep reading of poets routinely produce styles that detectors mislabel. Third, the broader discussion about AI's role in writing is not going away. Resources that help writers understand these tools - including what they can and cannot do - are increasingly relevant for those navigating both the craft and the controversy around it. For more on that evolving landscape, see AI for Writers.
Your membership also unlocks:
