Booksellers suspect AI firms are buying and then destroying rare books (arstechnica.com)

Recent concerns have arisen among booksellers that AI companies may be purchasing rare books and subsequently destroying them in the process of training their models. This alarming trend highlights the lengths to which tech companies might go to acquire high-quality long-form texts, often leading to the irreversible loss of valuable literary works. As the demand for vast amounts of data increases, the practice of quickly scanning and discarding these physical books is viewed as both disheartening and destructive to cultural heritage. The significance of this issue extends beyond just the loss of rare texts; it raises ethical questions about the methods employed in AI training. Notably, Google developed a non-destructive book-scanning technology back in 2009, which could provide a viable alternative for AI firms if they opted to invest in a more careful approach. However, this method is not without its flaws, such as potential text distortion and missed pages, which may deter companies focused on cost-efficiency. The situation highlights the tension between rapid technological advancement and the need for responsible stewardship of cultural resources, prompting discussions about how the AI/ML community can balance innovation with respect for historical artifacts.