Amazon’s AI Hunger Is Putting Rare Books Through the Shredder

Amazon’s AI Hunger Is Putting Rare Books Through the Shredder
Amazon built its name on selling books. Now it faces allegations that its race for AI training material is turning scarce physical volumes into disposable data.

The trail began last month, when 404 Media reported that AI companies were buying used books in bulk for training data. The outlet later placed a tracker in a shipment of rare books and found that it reached Amazon’s VGT3 facility in Las Vegas.

Workers at that site described a blunt, industrial process: large book shipments arrive, bindings are cut away so pages can be scanned faster, and the printed volume is destroyed. “The printed book is destroyed in the process.” The allegation is especially striking because rare and out-of-print titles may contain text unavailable online — precisely the kind of material AI developers increasingly need.

Amazon’s position is narrower. It told 404 Media that it “purchases books through commercial channels to improve the products and services customers use.” But the reporting says the company is buying rare books in quantity, cutting off their spines and digitizing them, raising a collision between the demand for clean training data and the preservation of physical culture.

The pressure is not merely about volume. Older books offer text that predates the current flood of AI-written material; training models too heavily on synthetic output can degrade their results, a risk often called model collapse.

The dispute also fits a wider argument over what AI becomes in human hands. One recent reflection compares the technology to dogs — a powerful companion whose superior capabilities once gave humans an evolutionary edge. Amazon’s critics see a harsher version of that bargain: a tool meant to expand knowledge consuming the artifacts that hold it.

Continue reading https://foxvector.com/stories/01a029e0-5513-1b64-7328-13e5d8e7368a