Amazon Scans & Shreds Books for AI Training Data

Amazon is reportedly engaged in a practice of mass-scanning and then shredding physical books to procure data for training its artificial intelligence models. This strategy reveals the intense demand for high-quality content to fuel the development of large language models.

The Scanning Process Uncovered

An investigation revealed how Amazon obtains this crucial data, tracking a rare book using an Apple AirTag.

• The tracked book ultimately led to an Amazon facility in Las Vegas.
• There, employees were observed systematically cutting the bindings of books before scanning them.
• This process effectively destroys the original physical copies after their digital data extraction.

AI’s Data Imperative

This extensive process directly addresses a significant demand from AI companies for high-quality, human-written content. Such material is particularly sought after when it involves older printed books not widely available online.

Leveraging these physical texts is critical for developing robust large language models, simultaneously mitigating the inherent issues often associated with AI-generated training data.

Corporate Strategy and Future Moves

Amazon acknowledged its practice of purchasing books through commercial channels, stating the intent is to enhance its services.

However, the company notably refrained from clarifying the specific reasons behind the physical destruction of these original copies. This strategic move aligns with Amazon’s broader efforts, as it is also actively developing its own artificial intelligence models under the ‘Nova’ brand, signalling a clear push into the AI domain.