Introduction
Artificial intelligence needs data to operate. Lots of data. For some companies, this means destroying rare books. According to a report by HedgieMarkets, AI companies are bulk-buying rare books, scanning them with high-speed machines, and then destroying the originals.
The Process
A company called ISBNdb facilitates these massive purchases. It allows orders of up to a million books while maintaining the anonymity of buyers. Pre-2022 books are particularly sought after because they are free from AI-generated text, making them ideal for training new models.
The Legislation
A federal judge ruled that this practice constitutes "fair use," as eliminating the original means only one copy exists at a time. This has paved the way for an acceleration of this activity, despite ethical concerns.
The Consequences
The destruction of rare books poses irreversible problems. Unlike a website that can be re-uploaded or a bestseller that can be reprinted, a destroyed rare book cannot be replaced. This is particularly concerning for works with few surviving copies, such as certain 18th-century botanical texts.
A Controversial Business Model
ISBNdb has built a business model around this process, offering NDAs to keep these purchases secret. They call it "digital preservation," but the reality is that unique pieces of our cultural heritage are being destroyed to improve marketing models or chatbots.
Industry Reactions
Companies like Anthropic, which hired the former head of Google Books partnerships, seek to acquire "all the books in the world." This raises questions about cultural impact and the responsibility of AI companies.
Conclusion
AI companies must find a balance between innovation and cultural heritage preservation. Destroying rare books for AI model training is a slippery slope that could have long-term consequences on our cultural heritage.
Let's discuss your project in 15 minutes.