This summer, sellers of rare books reported to sources about a surge in large orders for antiquarian texts. To uncover the buyers behind these transactions, investigators cleverly placed an AirTag inside one of the rare volumes. This investigation ultimately traced the shipments back to an Amazon warehouse located in Las Vegas. Each Amazon facility is identified by a unique symbol, and at the LAS8 site, the logo for the VGT3 team features a menacing T. rex holding a book. Employees from the VGT3 team have shared on worker forums that their warehouse processes substantial deliveries of printed books. The routine involves scanning the barcodes, removing the bindings, and scanning the individual pages, which results in the complete destruction of the physical books.
Why are AI companies interested in these old texts? Antique books are particularly appealing to AI developers because they contain unique content that is often absent from online databases and dates back to a time before the rise of large language models (LLMs). A bookseller who assisted with the investigation expressed, "Value comes in various forms. Beyond monetary worth, there's historical significance, intellectual importance, and even emotional attachment. However, AI companies disregard all these aspects; they merely seek the text as a collection of words." In response to inquiries, Amazon stated that it "acquires books through commercial channels to enhance and refine the products and services available to our customers."
What began as an online bookstore in 1994 has evolved over three decades into a company that not only purchases books but also obliterates them in the quest to convert them into data. The ethical question remains: is it justifiable to destroy books for the purpose of training artificial intelligence?
Informational material. 18+.
