xyz The pixel data needs to be tagged in all sorts of ways.
Fortunately for the language models, humans are quite willing to do that for free, in our spare time -- as the existence of social media shows. Most of us don't even care how someone might be using our text.
Tomcat Right, but the spent days will go to one person, and the hours saved will go to many, resulting in a total gain of years.
That's the best case, but I never thought altruistically when I started banging together a utility, and most of them were forgotten before they ever got used again. If you've got a geeky manager, they'll excuse any delay when you show them a neat piece of code. 🙂
Erich_L I assume that no, you don't/wouldn't feel any guilt at all even if starting with a "protected" image.
I probably would feel guilty if I wasn't buffered from the original by the software, but I'd get over it. Anyone who is really worried about it could check this out.
https://arstechnica.com/information-technology/2023/03/ethical-ai-art-generation-adobe-firefly-may-be-the-answer/
Copyright doesn't work in the digital age. We've been trying to shoehorn it in for decades, but it's what programmers refer to as a kludge. It might be somewhat functional now, but you know it will break as soon as anything changes. The only reason the concept ever existed is the pre-industrial idea of patronage -- in this case, society would go out of our way to encourage people to pay the author.
It still exists because wealthy people bought lots of authors' work and want to continue to charge money for them. (*cough* mickey mouse) Make no mistake, the people with the power to make decisions don't care about struggling artists, except in the sense that a farmer cares about a potato plant. However, they will probably attempt to quash learning systems to protect their own portfolio.
I'd normally say that such an effort is impossible -- the genie's out of the bottle, and it's not going back in -- but what if some hack puts together a legal "AI spotter". The ultimate copyright troll enforcer that tags millions of "offenders" every day. Then someone creates a framework of law that allows "AI judges" to rule on very limited cases of copyright infringement and redirect any attempts at payment to the copyright owner. (Sound familiar?) If you've read Melancholy Elephants, this is worse.
I can think of much, much nastier things that this software could be (and probably is being) used for, but I don't want to give anyone nightmares tonight. 🙂 Anyway, it's more likely that the results of AI art will be declared original, since the money makers would be happy to dispense with the bothersome human artists and save a few bucks.