AI companies are shredding rare books

This is basically AI slander. They are not destroying rare books, certainly not "18th century botanical" treatises. Anything pre-1929 is public domain anyway. And nobody is going to boy or sell a truly rare museum or collector grade book as part of some bulk lot. What they are purchasing are common and relatively obscure books, magazines, etc, which have been languishing unsold, unread, and unknown in used book warehouses. They would likely have eventually rotted and disintegrate in obscurity anyway. This scanning process actually preserves them. And even destructively it is highly unlikely that the few such copies collected and scanned by AI companies will literally represent that last copies of any book. Or, in the extremely rare case where such is true, those books are rare, and ended up in bulk lots, precisely because they were never valued. In any event, all it will take is an agreement with the copyright holders to make these digital copies available to the public. And in the meantime the texts will influence the weights in AI models, with the "gist" preserved rather than exact quotation.