00:00Are AI companies destroying rare books?
00:02That's a claim you see all over social media right now.
00:05Here's why reality is more nuanced.
00:07Yes, Anthropic is buying thousands of physical books worldwide.
00:11The reason?
00:11AI training data.
00:13AI companies have already trained their models on vast amounts of text scraped from the internet.
00:18But fresh, high-quality human-written data is becoming harder to find.
00:21And much of today's newly published content may already contain AI-generated material.
00:26That's why they're turning to books.
00:27Especially those that haven't been digitized and were written before the AI era.
00:31But what books aren't already online?
00:33Well, the ones that attract little interest.
00:36Old travel guides or super niche technical books.
00:38Not very exciting for most readers, but valuable for AI.
00:41They're curated, reviewed, and structured.
00:44And yes, the books are destroyed during the scanning process.
00:47But it's unlikely that these are rare cultural treasures disappearing from library shelves.
00:51In most cases, they're low-demand books that exist in multiple copies.
00:55Critics, however, warn that at the scale of millions of books, no one can verify every copy.
01:00And that's what worries them.
01:01A rare print might slip through.
01:02A rare print might've been published through.
01:03A rare print might have been published in a series, by investing in a series, by a month of
Comments