Quote:
Originally Posted by redwhiteandblue
But.....what is the difference between a human being studying an artist's work and then copying the style himself, and a human being showing that work to a computer algorithm and training it to do the same? One is okay, the other, not?
|
I completely get your point... But it also works the other way around.
If some artist's work ends up looking too similar (think trademarked characters), there's a chance that the artist might end up getting sued.
But what if, for example, the AI randomly ends up producing a character that looks suspiciously similar to whatever popular trademarked character or copyrighted art work?
Seems a bit like a grey area at this point. Sure, they can force someone to take it down, but since there's no actual person really to blame here, one might decide to go after the team behind the AI.
The big issue here could be the way they've scraped the data set... which again, is copyrighted material and they've scraped it without consent, which is illegal. Might not sound like that big of a deal, but that could become a serious issue once that (illegally) scraped dataset is used in a commercial way.
What if Nintendo decides to sue that AI company because of all the Mario-looking mothafuckas that got spit out by the AI, I think they've already got a case by simply stating that their trademarked characters never even belonged to be in that dataset to begin with, since it's used in a commercial way.
Which kinda explains why this certain "big company", which has always been known for being against open sourcing code, suddenly felt the urge to purchase the largest open source platform on the internet... They needed a large dataset for their AI, and they needed to obtain it in a legal way (it's in the terms of service).