8 ms·
No one will be able to prove the source of the data - was it Disney content, or was it someone copying the animation style of Disney content?
by TobyTheDog123 4y ago
No one will be able to prove the source of the data - was it Disney content, or was it someone copying the animation style of Disney content?
- ChatGTP 4y agoEventually you will need to show sources and I’m sure we’ll have ways to extract sources from the models. ma these things are not magic after all.
- amanaplanacanal 4y agoI’m not sure why sources matter at all. In copyright you compare the end result, not how you got there.
- blibble 4y agothis isn't true at all the process the work was generated through is of primary importance for copyright the existence of the concept of "derivative work" should make this obvious see the reverse engineering of the IBM BIOS for another example
- CuriouslyC 4y agoThere are specific criterion for fair use: https://fairuse.stanford.edu/overview/fair-use/four-factors/ https://fairuse.stanford.edu/overview/fair-use/four-factors/. None of them are process specific. Maybe bone up on the law rather than getting yourself in any deeper.
- blibble 4y ago> Maybe bone up on the law rather than getting yourself in any deeper. we're not talking about fair use here buddy
- CuriouslyC 4y agoAnd that's probably why you're talking to yourself.
- blibble 4y agoMaybe try reading the thread rather than getting yourself in any deeper.
- CuriouslyC 4y agoYou're the one trying to ignore fair use as it has been broadly applied to creative works for something like 150 years, and make this about a very narrow slice of software copyright law even though people in the thread are talking about images and music. You're being willfully obtuse to try and make an argument because you are emotional about AI creative output.
- blibble 4y agoanother strawman
- kmeisthax 4y agoYou actually can, but only for things the model memorized: https://arxiv.org/pdf/2301.13188.pdf https://arxiv.org/pdf/2301.13188.pdf
- bryanrasmussen 4y agoI've noticed programmers often think the law will be worked around by 'tricks' your honor, this LLM produces content that looks like various Disney intellectual products, it was obviously trained on that data. your honor, no we trained it on a lot artwork that was copying the Disney style! Judge: do you have this corpus of data you trained it on for the court to inspect. uh, no. summary judgement for plaintiff.
- newswasboring 4y agoWhy will this lead to summary judgement to the plaintiff, neither side has presented concrete proofs, just assertions. Serious question.
- maxbond 4y agoThe standard in US civil court is preponderance of evidence, not reasonable doubt. You'd have to go further then making an assertion (by bringing in an expert witness or something) but you don't necessarily have to prove something definitively. You just have to make the more compelling argument - you have to present evidence that it's more likely that your position is correct. Since the hypothetical AI company wasn't able to produce anything in it's defense, it lost. IANAL
- dragonwriter 4y ago> you don't have to prove anything, strictly speaking. Strictly speaking, you do have to prove things (if you have the burden of proof on the specific issue), and the standard of proof is usually “preponderance of the evidence” (though there are a few other standards that apply to particular issues/circumstances in the civil justice system.) “Beyond a reasonable doubt” is a different standard of proof used for conviction in the criminal justice system, but it is strictly not correct to call proof under other standards something other than proof.
- maxbond 4y agoFor sure, I was responding to the idea of "concrete proof" and why it didn't need to be "concrete proof," but I'm sure you're right. Thanks for keeping me honest. I've edited my comment to reflect your correction.