AI's Hidden Cost: Rare Books Face Destruction in Training

The pursuit of advanced AI is driving a disturbing trend: the wholesale destruction of rare and valuable books. Companies, including those behind ChatGPT, are acquiring vast quantities of physical books to feed their large language models, often by tearing them apart and scanning the pages.

A Secret Operation: ‘Project Panama’

Operation ‘Panama,’ conducted by Anthropic, exemplifies this practice. Companies are efficiently stripping books of their covers and bindings, feeding the loose pages into scanners – a cheap and rapid method for data ingestion. This process, however, results in the irreversible loss of these historical texts.

Rising Concerns and Book Seller Resistance

Librarians, independent booksellers, and book lovers are voicing strong objections. Kennys Bookshop in Ireland recently flagged a suspicious order for 5,000 obscure titles, raising concerns about the potential destruction of irreplaceable works. Owners are now refusing such orders, fearing permanent loss.

The Dystopian Reality

While not involving literal burning, the impact mirrors the concerns raised in dystopian literature like Fahrenheit 451. The question arises: is the drive for AI innovation worth sacrificing a tangible link to our cultural heritage?

A Growing Backlash

The practice is fueling suspicion within the book selling community, as orders are placed for books that subsequently vanish. The race to train AI models is creating a silent crisis for the world’s literary past.