Author Steven Rosenbaum says AI led to incorrect quotes in his book

In short: Author Steven Rosenbaum says AI tools he used for research contributed to a few incorrect or made-up quotes in his new book, and he is now reviewing citations.

What happened

Steven Rosenbaum recently published The Future of Truth: How AI Reshapes Reality, a book about how AI can blur what is true. A New York Times investigation found what Rosenbaum later acknowledged were “a handful” of quotes in the book that were either wrongly attributed or “synthetic,” meaning they appear to have no real source.

One quote was attributed to tech reporter Kara Swisher, who told the Times she never said it. Another was attributed to professor Lisa Feldman Barrett, who said the quote was not in her work and was wrong.

Rosenbaum told Ars Technica he used AI tools like ChatGPT and Claude during research to find articles, summarize themes, and suggest people or papers to look up. He said the writing, interviews, and arguments were his, and that AI did not write the book. He also said AI-sourced notes were marked in his files and then sent through a fact-checker and copy editors, but some errors still slipped through.

Rosenbaum says he is now working with editors on a full “citation audit,” which is a line-by-line check of sources, to correct future editions. Even after this experience, he said he does not plan to stop using AI for research because it saves time, even though it can be convincingly wrong (like a fast assistant who sometimes makes things up).

Why it matters

This case shows a simple problem: AI can produce text that looks confident but is not true. For readers, it is a reminder to treat quotes and sources carefully, especially when AI tools were used anywhere in the research process.

Source: Arstechnica