News Novelists Sue OpenAI for Copyright Infringement Over Books Used as Training Data

Why is it ok for a person to read something and learn, and then make money off what he learned; but it's not ok when an AI does it?
I'm guessing the distinction being that you learn stuff for yourself and subsequent application of that knowledge is limited by you being human while an AI can learn stuff once and apply the knowledge worldwide almost instantaneously.
 
I'm guessing the distinction being that you learn stuff for yourself and subsequent application of that knowledge is limited by you being human while an AI can learn stuff once and apply the knowledge worldwide almost instantaneously.
Only to "provide plot summaries and answer questions about the books", as the plaintiffs themselves state, not to share the actual texts in violation of copyright law.

To me, this looks more like a cheap PR move to promote their crappy books (the first I checked had 3.05 rating on GoodReads) than an serious argument for copyright infringement.
 
  • Like
Reactions: bit_user
I'm guessing the distinction being that you learn stuff for yourself and subsequent application of that knowledge is limited by you being human while an AI can learn stuff once and apply the knowledge worldwide almost instantaneously.
That's not a violation of the copyright law they're citing, though. They're claiming ChatGPT is a derivative work, but there's a legal standard for that and I think it's one that ChatGPT won't meet.

Now, if they could trick ChatGPT into regurgitating passages of the books, verbatim, they might have a real copyright case. However, I'm guessing they tried and it didn't or wouldn't.

I think the plaintiffs are essentially playing the lottery. Even though they probably know their chance of winning is small, they figure the payout could be big enough to make it worth a shot.
 
  • Like
Reactions: BX4096