The French literary world is abuzz with controversy surrounding Canadian-Haitian writer Thelyson Orelien's best-selling debut novel. An X account sparked the debate by claiming that the AI-detecting program Pangram concluded that the tale of a Haitian migrant traveling to Canada was largely written by a text-generator. Orelien denies the allegations, stating he will prove that he wrote the book, which has been sold to publishers in over 20 countries.
Orelien expressed his sentiments, saying, “I didn’t write this novel with my head, but with my guts and my heart,” in a recent interview with AFP. To investigate the claims, AFP ran excerpts of the book through various AI detection tools, yielding a range of results from "very likely" to have been written by a human to "100 percent" AI-generated. This disparity raises questions about the effectiveness of these detection tools.
Dozens of tools are now available online to detect AI-generated text, primarily designed for use in education or publishing. These tools typically have large collections of AI-generated text alongside human-written content, which they compare to identify potential AI-generated writing. Researchers at the University of Chicago found that Pangram achieved near-zero error rates on medium-to-long extracts in a paper published in October 2025.
According to Thierry Poibeau, a researcher at France's national CNRS research institute, the programs look for "linguistic tics" such as triadic phrasing or em dashes. These telltale signs were common in generated texts because the tools were trained on American scientific writing or American novels. Pangram claims it can detect AI text by analyzing subtle cues in the writing, including repetitive words or phrases and monotonous sentence structures.
However, experts question the reliability of these tools. Many publish percentage figures about their accuracy, but these figures are unverifiable and do not account for the risk of false positives. Pangram CEO Max Spero stressed that the tool should not be the ultimate arbiter but rather a starting point for further investigation.
The detection systems also face a "moving target problem" as AI programs constantly evolve. Thierry Poibeau points out that the tool may identify something particular in the writing, but that might be a writer's style. Orelien claimed that his style is based on Haitian and Caribbean traditions, which may not be recognizable to the software.
Other experts have weighed in on the issue. Benoit Raphael, an entrepreneur and writer specializing in AI, used Pangram to test Orelien's work and obtained results indicating a high likelihood of AI-generated content. Some critics had already raised doubts about the book's authorship due to its repetitive sentence structure and overuse of analogies, hallmarks of AI-generated writing.
Key points
- The controversy surrounding Orelien's novel has sparked debate about the reliability of AI writing detectors.
- AI detection tools have improved significantly in recent years but still face challenges in accurately identifying AI-generated content.
- The reliability of AI writing detectors is crucial in maintaining the integrity of literary works and preventing potential misattributions.
