I published a book written with AI where every claim carries a truth label, and rival AI models reviewed it under instruction to be ruthless

An author published an AI-written book where every claim includes a truth label, and rival AI models reviewed it ruthlessly. The author notes that AI-generated books often blend real physics with speculation, making it hard for readers to distinguish facts. To address credibility issues, the book incorporates corrections for past errors, such as a p-value discrepancy and an inaccurate word count. The author suggests that this "marker system," which cost nothing but discipline, is crucial for maintaining trust in any domain where LLMs are used.