The price paid by a journalist for using AI after publishing quotes invented by chatGPT: “Generates compelling quotes that you're tempted to use”
A top European journalist has lost his credibility and been suspended after it emerged that he used artificial intelligence to generate misquotes in his articles.
Peter Vandermeersch admitted his mistake. PHOTO: Mediahuis
Publishing group Mediahuis has decided to suspend one of its senior journalists, Peter Vandermeersch, after an investigation revealed, and he later admitted, that he had used artificial intelligence to “wrongly put words in other people's mouths”.
Peter Vandermeersch, former Irish editor of Mediahuis, confessed that he “fell into hallucinations” (a term used for AI-generated errors) while using the technology. He explained that he used AI, including Google's ChatGPT, Perplexity, and NotebookLM, to summarize articles, and that he did not verify that the quotes in those summaries were accurate before publishing them in his Substack newsletter.
The AI-made quotes were discovered by the publication NRC, part of Mediahuis and where Peter Vandermeersch had been editor-in-chief in the 2010s. According to it, the journalist published “dozens” of fake quotes, with at least seven people quoted in his posts declaring that they did not make the claims attributed to them, according to The Guardian.
“I mistakenly put words in people's mouths when I should have presented them as paraphrases. In some cases, they reflected my interpretation of what they said. It wasn't just careless, it was wrong.” Peter Vandermeersch admitted.
“It's all the more painful that I made the exact mistake I warned my colleagues about: these language patterns are so good that they generate irresistible quotes that you're tempted to use as an author. Of course, I should have checked them. The necessary human oversight, which I constantly advocate, was lacking.” he continued.
The irony of the situation is all the greater as, on his blog, Press and Democracy, the journalist frequently writes about “the vital link between a free press and a healthy democracy.”
The official press release of the Mediahuis group
Using artificial intelligence programs to write his articles cost Peter Vandermeersch not only suspension, but also loss of credibility.
“At Mediahuis we apply strict rules regarding the use of artificial intelligence, where rigour, human oversight and transparency are essential. The fact that these principles were not respected goes against our standards and commitment to readers that we support reliable journalism. We are discussing this case with Peter Vandermeersch and have decided to temporarily suspend him.” conveyed, in a statement, the executive director of Mediahuis, Gert Ysebaert.
Mediahuis also removed a number of articles written by Vandermeersch from the Irish Independent website.
AI tools like ChatGPT are used by hundreds of millions of people for various tasks, from simple recipe recommendations or small home repairs, to complex academic research, but in some cases they make major mistakes and cannot replace humans.
“Journalism is human work. I remain convinced that AI can be a powerful tool, one that can help journalism become better, more in-depth and more accurate. But not using AI in the way I did in the first few months of this blog,” Peter Vandermeersch said, declining to comment further.
