> Except they’re not storing copies of any of these. How do you think gtp(n+1) is trained? We literally have people in the AI field admitting part of their tr

> Except they’re not storing copies of any of these. How do you think gtp(n+1) is trained? We literally have people in the AI field admitting part of their training dataset is a pile of torrented books.