We Asked The Author Of ‘The Future Of Truth’ To Explain How He Used AI. It didn’t go well

Earlier this month, WIRED published a new book by Steve Rosenbaum, The Future of Truthwhat it looks like artificial intelligence it distorts people’s perception of reality. Soon after, The New York Times report that the book contains more than half a dozen words made up or mispronounced. In his statement, Rosenbaum, who has a master’s degree in “truth” from New York University, admitted that he accidentally included “a small number” of “false or fabricated” statements. Ironically, the accuracy of a book about AI’s impact on reality has now come under intense scrutiny because of its author’s use of AI.

After the Times story, WIRED reviewed our 1,450 words. A fact-checking team had reviewed it before publication, and we confirmed that the wording and facts were correct. But WIRED’s AI editorial policy prohibits the publication of AI-generated and AI-edited articles, and a reader email that referred to itself as “AI-written” raised some questions about Rosenbaum’s use of AI tools. In The Future of Truth‘s acceptance section, Rosenbaum writes that ChatGPT, Claude, NaturalReaders, ProWritingAid, and Grammarly helped “refine and polish the presentation of (his) ideas.” What does that really mean?

WIRED ran the text through several AI recognition tools, including Pangram, GPTZero, and ZeroGPT. Each application stated that it was either created by AI, or created by an AI with high confidence. But AI detection tools are imperfect, and can return incorrect readings. So WIRED’s head of research emailed Rosenbaum to ask him how he had used AI to write the article.

He wrote: “Like many authors working today, I used AI tools in the research and development phase of the book, including source discovery, reasoning, voice response, and language processing.” But, he emphasized, “opinions, reporting, arguments, and final authors are my own, and the WIRED content was not generated by AI and was simply published as is.” He urged WIRED editors to be cautious about relying on AI detection tools, noting that mistakes can happen.

At this point, the senior editors at WIRED asked me to look at the piece, because I had written about it The decline of AI in his own variety appearance starting in 2024. My first step was to run all the text in the book through the Pangram recognition tool. (Although all AI-detection tools have limitations, and can show errors, Pangram is the current gold standard.) It came back that the book seems to be 53 percent of AI generated by AI, and an additional 9 percent registered as AI-assisted.

I called Rosenbaum and asked him to explain in detail how he used AI to write the book, and whether he disputed Pangram’s results. (BenBella Books, whose articles were published The Future of Truthdid not return requests for comment. Simon & Schuster, which distributes BenBella’s books in the United States, declined to comment.)

Rosenbaum would not quantify the accuracy of Pangram’s results. In fact, he didn’t want to talk about them. He said: “I will not take part in that discussion. “It is like saying, do you beat your wife? It is one of the reasons why there is no answer. “

He offered, instead, to clarify his writing process. He says that at the beginning of the manuscript, he used AI tools such as search engines, which helped him to find information about the research-heavy sections of the book. To demonstrate how to do this, he asked ChatGPT to explain it to me, then read it out loud. AI search made clear my previous articles, including work on AI-generated “zombie media sites”.