Amazon, the company that built its empire by selling books, is now destroying rare texts to train its artificial intelligence models. The revelation has sparked outrage among booksellers, historians, and readers who see the practice as an irreversible loss of cultural heritage.
Why Rare Books Are Valuable for AI Training
Rare books hold unique value for training large language models (LLMs). According to Futurism, these models have already trained on whatever is available online. Rare texts offer fresh, unpublished content that cannot be found elsewhere, making them incredibly valuable for improving AI's knowledge and capabilities.
The process involves buying antique books, scanning or ingesting their contents, and then destroying the physical copies. This practice has been described as happening at an "incredible scale," raising serious concerns about the loss of irreplaceable works.
Booksellers Raise the Alarm
Booksellers are among the loudest voices condemning the practice. According to The Guardian, Australian booksellers have raised alarms over the "horrific" destruction of rare titles to feed AI supply chains. They describe these books as "more than just objects," emphasizing their cultural and historical significance.
"More than just objects" — The Guardian
The concern is not just about losing physical books. It is about losing the knowledge, history, and context they carry. Once destroyed, these texts cannot be recovered, and their contents exist only in the AI models that consumed them.
The Scale of the Problem
Reports indicate that AI companies are buying and destroying millions of physical books to train their models. According to YouTube coverage, this practice is designed to facilitate regurgitation and feed AI chatbots with unique content. The scale of destruction has been described as "incredible," with no signs of slowing down.
- Rare books are targeted because they contain content not available online
- Physical copies are destroyed after their contents are ingested
- The practice is happening at an "incredible scale"
- Booksellers describe the destruction as "horrific"
Our Take: The Cost of Progress
To put it plainly, this is a troubling trade-off. Amazon and other AI companies are prioritizing technological advancement over the preservation of cultural heritage. While training AI on rare texts may improve their capabilities, the destruction of physical books is a loss that cannot be undone.
In our view, there is a better way. Companies could digitize these texts and preserve the physical copies in archives or libraries. Destroying them is not just wasteful — it is reckless. Readers and historians should pay attention to this issue because it affects all of us. Once these books are gone, they are gone forever, and future generations will have no access to them.
The question is not whether AI should learn from rare texts. The question is whether we are willing to sacrifice our cultural heritage for that knowledge. We believe the answer should be no.
