I am genuinely trying to get better at art. I’m not there yet (likely never will be), the lying machine is still better than me.
The context:
This is my sketch.
And this is what the ai output.
I like to think I poured my heart and soul into it. I know there are people who will tell me that I’m terrible for using ai at all. I’m also sorry if this is the wrong community to ask this question (ask reddit would delete my post instantly if I tried to post there).
Again, is this slop? I am not an artist. I drive a forklift real good, that’s my skillset. So if I were to use the ai upscaled version for my book, well, I’m asking for opinions.
Personally I think the same argument pirates use with content can be used here. Usage of the art generators != a sale for the artist. I don’t agree with artists not being compensated either piracy is piracy regardless if its a big company or an artist.
However, It was already a stretch to claim that pirating a digital media that is 1:1 is equal to losing a sale, but it’s even even bigger stretch to claim that an image that is generated by a generator using training data of a bunch of artists and images pushed together would equal a sale. In most cases you wouldn’t even be able to identify the artist in the first place(or even know they existed).
As for driving the artists out of the market, I personally don’t believe it ever will manage that, because of how it works. It needs something to train off of, when artists decide to leave the market nothing new will be able to be trained, which will eventually kill off the ability to train them further, which will leave them either to use stale training data, or use existing models to train themselves, which have been pretty spotty in quality in general.
Plus, at the end of the day, art is a profession for many and a personal hobby for even more. I don’t think it’s going anywhere any time soon, regardless of tools available, I think I would say the same with LLM story-wrights as well, for any new field breaking genres or concepts, a humans touch is needed or else it will just piece together what it already knows.
Being said, hopefully some point the fields start actually paying for the content they are using, I expect it will eventually happen, its just legal frameworks are slow. (ignoring the few models out there that do pay for content like others have said such as Adobe)
Honestly my biggest concern is going to be what happens when those AI companies purchase data off of media websites who claim ownership of all data submitted. Usually they operate off a shared copyright where by submitting the content you allow them to use it how they see fit but you as the artist keep ownership. For example Deviantart while claiming they have no ownership also has this in their policy.
Which grants royalty free permission to redistribute how they please.