AI put "synthetic quotes" in his book. But this author wants to keep using it.

Journalist and author Steven Rosenbaum faced controversy when a New York Times investigation revealed "synthetic quotes" in his new book, "The Future of Truth: How AI Reshapes Reality." Rosenbaum acknowledges that his use of AI research tools led to improperly attributed or fabricated quotes, some of which were explicitly denied by the individuals to whom they were attributed. He is currently working on a "citation audit" to correct future editions of the book. Despite this setback, Rosenbaum remains committed to using AI in his research and writing process. He describes AI as "magical" for its ability to connect ideas and provide new avenues for thought that he wouldn't discover on his own. He maintains that while AI assists with surfacing ideas and summarizing themes, the actual reporting, narrative, interviews, and conclusions in his book are entirely his own. Rosenbaum clarifies that he used tools like OpenAI's ChatGPT and Anthropic's Claude to gather information, tagging any AI-generated notes with a warning to distinguish them. These notes were then passed to a fact-checker and two copy editors. Out of 285 citations, six were identified as problematic by the Times, including three "synthetic quotes" with no apparent source. The incident highlights a critical challenge for traditional fact-checking, which is often ill-equipped to verify AI-assisted research. The assumption that quotes from cited works are directly copied is invalidated when AI tools are involved, necessitating an additional layer of skepticism. This issue is compounded by financial pressures on newsrooms and publishers, leading to cuts in fact-checking and copy-editing departments. Rosenbaum agrees that publishers need new verification workflows tailored for the AI era, including mandatory source tracing and clearer standards for AI-assisted research. He emphasizes that he did not intend to fabricate anything, but rather, the AI-generated information, which appeared authoritative, unfortunately made its way into the book. This incident reflects a growing trend of AI-generated errors in various fields, underscoring the urgent need for enhanced scrutiny and robust verification processes.