Nvidia denies pirate e-book sites are "shadow libraries" to shut down lawsuit

I have another take on this dispute. We permit copyright because we reckoned it incentivizes a social good. AI is a bigger social good. An additional condition of copyright should be to submit your text to the pool of text available to AI. AI is that important, that revolutionary, that beneficial.

> "Nvidia [...] denies that hosting data in or distributing data from the data repositories necessarily violates the US Copyright Act," Nvidia's court filing said.

What's allegedly claimed seems extremely strange since downloading copyright works without having obtained a license has been prosecuted very actively.

I don't see how there's any legal uncertainty at all here. If you own a book you can scan it and train AI on it, but if you don't, you can't legally download it even if you're only training AI on it.

Simply, if this is permitted there's no copyright. After all, this is a matter of commercial activity, not just some person downloading a book which he can't afford and is curious about.