with ai, comes lawsuits

[me just saying]

Been reading about the different lawsuits and the threats of lawsuits over the use of online data to train the different ai packages. I was just wondering how is this so different from google sending out spiders to collect the very same data for their search results? What do you say?

 

[iamthecosmicdust]

Web crawling for search engines operates within the bounds of publicly accessible content and user consent mechanisms, whereas AI training data can involve more nuanced privacy considerations and legal implications.

 

[Mooncatt]

It'll be interesting how this plays out in the courts, but here's my personal take.

I'm in an art focused fandom and this sort of topic comes up often. One of the most common arguments against AI is that it's theft.

I say it isn't.

Compare AI training to any human artist in training. In both cases, the trainee uses existing works to learn from. It could be studying shading methods in imagery, or world building in novels. In both cases, the trainee uses those references to help build up their own style, and styles can not be copyrighted. The argument goes that the AI training is stealing the art is referencing, but those people have no problem with an artist doing so, even when the end product is virtually the same.

Yes, it's possible for an AI to make a copy of an existing work, but that's something that has been possible forever. In this case, it's not the AI that is at fault, but the person behind the work that intentionally created the copied or misleading work. Those cases should be prosecuted where applicable, but not because it was created with AI. Existing laws on copyrights would apply the same as if that same person created the copy by hand.

It's also worth noting that work created with generative AI can not be copyrighted, at least in the U.S. There have been attempts at doing so, and the copyright office has denied all of them.