Physical Address
304 North Cardinal St.
Dorchester Center, MA 02124
Earlier this month, WIRED published an excerpt from Steve Rosenbaum’s intriguing new book, The future of truthwhich looks at how artificial intelligence It distorts people’s sense of reality. Shortly thereafter, The New York Times I mentioned The book contains more than six fabricated or misattributed quotations. In a statement, Rosenbaum, who has a master’s degree in “fact” from New York University, admitted that he had mistakenly included “a handful” of “incorrectly attributed or synthetic” quotes. In an ironic development, the veracity of a book about how AI affects reality is now under intense scrutiny because of how its author used AI.
After the Times story was published, WIRED took another look at our 1,450-word excerpt. Our fact-checking team reviewed it before publication, and we reconfirmed that its quotes and facts were accurate. But WIRED’s Generative AI editorial policy prohibits the publication of AI-generated and AI-edited writing, and a reader email describing the excerpt as “blatantly AI-written” raised further questions about the extent to which Rosenbaum used AI tools. in The future of truthIn the acknowledgments section, Rosenbaum wrote that ChatGPT, Claude, NaturalReaders, ProWritingAid, and Grammarly helped “refine and refine the presentation of his ideas.” What does that mean exactly?
WIRED ran the snippet through several AI detection services, including Pangram, GPTZero, and ZeroGPT. Each service indicated that it was either AI-generated, or AI-generated with high confidence. But AI detection tools are not infallible, and can return inaccurate readings. So WIRED’s head of research emailed Rosenbaum directly to ask if and how he had used AI to write the excerpt.
“Like many writers working today, I used AI tools during parts of the book’s editorial research and development process, including source discovery, brainstorming, structural feedback, and language refinement,” he wrote in response. But he stressed that “the ideas, reports, arguments and final authorship are mine, and WIRED excerpts are not generated by artificial intelligence and then simply published as is.” WIRED editors urged caution when trusting AI detection tools, noting that false positives can occur.
At this point, WIRED’s senior editors asked me to take a look at the episode, because I covered it AI slope In it diverse Forms Since 2024. My first step was to run the entire text of the book through Pangram’s detection tool. (While all AI detection tools have limitations, and can show false positives, Pangram is the current gold standard.) The book was shown to appear to be AI-generated by 53%, with an additional 9% scored as potentially AI-powered.
I contacted Rosenbaum and asked him for a more detailed description of how he used artificial intelligence to write the book, and whether he questioned Pangram’s findings. (Ben Bella wrote, published by her imprint The future of truthHe did not return requests for comment. Simon & Schuster, which distributes Benbella’s books in the United States, declined to comment.)
Rosenbaum did not comment on the accuracy of the Pangram results. In fact, he didn’t want to talk about them at all. “I’m not participating in that conversation,” he said. “It’s like saying, Do you beat your wife? “It is one of those accusations to which there is no answer.”
He instead offered to explain extensively his editorial process. He says that at the beginning of the writing process, he used artificial intelligence tools as search engines, which helped him obtain information for sections of the book that required a lot of research. To demonstrate how to do this, he asked ChatGPT to describe me, then read the results out loud. AI research accurately described some of my previous stories, including work on AI-generated “zombie media sites.”
