AI Giants Frenziedly Snatching Rare Antique Books for Training Data; Clean Corpora Become the Scarcest Asset in the AI Era; A-Share 'Old Book Repositories' Directly Benefit

English summary

AI giants are frantically grabbing rare antique books as training data; clean corpora have become the scarcest asset in the AI era, directly benefiting the 'Big Three' A-share old book repositories! The overseas AI community is in an uproar—AI giants, in the race for clean training data, have begun snapping up rare antique books, cutting off spines for scanning and then shredding/destroying them. Even an 18th-century botanical unique copy with very few surviving copies was shredded. Even Musk couldn't stand it and urgently ordered his team to scan without damage. In a nutshell: internet data is now polluted by AI junk; future clean corpora are all locked in publishers' old book repositories. 1. China Science Publishing (中国科传) → A 70-year academic publisher under the Chinese Academy of Sciences, holding tens of thousands of journals and monographs in science, engineering, and medicine. Scientific and technological corpora are the most expensive data for AI training, and China's highest-quality 'antique-grade' scientific literature is locked in its journal library.…