203
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
this post was submitted on 20 Jan 2024
203 points (92.1% liked)
People Twitter
5228 readers
987 users here now
People tweeting stuff. We allow tweets from anyone.
RULES:
- Mark NSFW content.
- No doxxing people.
- Must be a tweet or similar
- No bullying or international politcs
- Be excellent to each other.
founded 1 year ago
MODERATORS
It's a little more nuanced as AI models keep spitting out verbatim training data when people figure out the right queries. E.g. the other day they banned you from asking chat GPT to repeat the same word forever as it'd do that for a while, then just spew out something it had been trained on. If someone reads thousands of articles on a subject, then writes the exact contents of one of them, that's definitely plagiarism.
There's also the issue that when a human reads a lot, they have to pay for a lot of books and view a lot of ads and pay taxes that fund a library system buying books, too. The human extracts value from what they've read and gives something to its author. Megacorporations training AI models are only extracting the value and aren't paying for the privilege.
It's not like I said it is because it IS usually like I said it is, except in intentional edge cases. Got it.
I said it was more nuanced than you said, which is pretty different to saying it's not like you said. There are big similarities to a human who's read lots of books, but the equivalent human is pirating all their media while being rich enough to pay for it, and sometimes passing off other people's writing as their own. Neither of those things are allowed when humans do them.