The question about the legal and moral aspects of training on works of other artists is related, but a different discussion.
I suppose it would depend on who the “artist” is considered to be at the end.
Say for instance I had an idea that I wanted a painting of Sir Issac Newton wearing a cowboy hat and riding a mechanical bull, and I commission a painter to create my vision. Instead of using paints or pencils or anything to create the image the painter goes online and downloads a bunch of pictures of Isaac Newton and mechanical bulls and collages them together in a way that looks kind of like an original painting.
Who is the artist in that case? It’s not me, since I didn’t make anything. It’s not the painter since they didn’t actually create anything original, they just stole a bunch of pictures someone else took. It’s not the people who made the original images that the painter stole since they never even agreed to be part of any of it.
We hit the same dilemma with AI. The person putting in the prompts hasn’t really “created” anything. The AI engine hasn’t created anything either, it just takes parts of other existing works. The people who made the original works had no say in any of how their work was used.
How is that “art”?
I love playing with AI to make silly images or even workshop ideas for things I might do in the future, but I wouldn’t call it “art”
I disagree with the premise that such mosaic of online pictures wouldn’t be “original” piece of art. It absolutely qualifies by my books
Who is the artist though?
One who wrote the prompt. It may be the AI that does all the heavy lifting but it’s still a tool and alone it doesn’t create anything.
But the person who wrote the prompts didn’t create anything. With AI there really is no “artist”.
How did they not create anything? They inserted a prompt into the tool and received a picture.
They had a rough idea and left it to the AI to make any sense of it and “create” something.
Painters can either splash paint on the canvas or spend months working on a photorealistic masterpiece. There’s absolutely a difference in skill needed for both but to claim the former is not art would also be gatekeeping.
That argument also disregards the actual difficulty of crafting the perfect prompt to get the AI to output what you want it to. Anyone can create pictures with it but it’s not trivial to get it to create exactly what you want.
I think you’re confusing art with « content ». When I experience art (of any kind, painting, music, writing, highbrow or lowbrow), I’m interacting with the artist, their intention, sensibility, politics, etc. I don’t feel that connection with statistically generated images or prompt engineers, sorry.
I do photography and I’ve heard people analyze my work and try and find some meaning, intention or a message that I’m trying to convey with it.
The reality is that I took 150 pictures and that was the one I liked the best. There’s nothing to it for me except how it looks. The fact that I managed to capture that specific photo is hardly anything but an accident. There is no meaning to it and whatever meaning one imagines seeing there is just in their mind. It’s a story you’re telling yourself and you’d come up with a similar story from a piece made by AI that you didn’t realize was such. If it stops being art at the moment you learn it was made by AI but you accept it as art when it was made by human even if it was, in fact, an accident, then that’s exactly the gatekeeping I’m talking about.
You took the pics and you chose the one you liked best. Enough said.
Tbf, that’s a lot of what making a good AI image is as well, messing with prompts and parameters then finding something you like.
Art allows you to experience the world through someone else’s senses and unconscious. The AI didn’t choose to take pictures of something. The AI doesn’t have subjectivity, which is what makes art interesting. If I find a box filled with pictures and find one I like, that does say something about my tastes, but it doesn’t remotely make me an artist.
But TBF, as an artist, I wouldn’t expect to connect with anybody who doesn’t care about the difference between human-made art and fancy statistical averages of troves of pillaged art.
Best thing I’ve ever read by far on the subject. Thought I’d share. https://www.newyorker.com/culture/the-weekend-essay/why-ai-isnt-going-to-make-art
And not every type of gatekeeping is bad.
The question about the legal and moral aspects of training on works of other artists is related, but a different discussion.
Thats not the main issue either. The issue is that Corpos rather prompt an LLM than pay for their artists.
And not every type of gatekeeping is bad.
Right?
“You must be this tall to ride” isn’t because the amusement park hates short people. It’s a safety issue.
I don’t see how that comparison translates to the topic about AI art.
If AI can create better content than humans can then people will rather consume that. I don’t see why you should artificially limit this. If someone thinks that AI content is not better then that’s who the audience is for the remaining human creators. AI can already create better looking photos than I can, but it has zero effect on my desire to do photography. I don’t see what the issue is.
If AI can create better content than humans can then people will rather consume that.
it can’t create better content though
I don’t see why you should artificially limit this.
LLMS limit themselves already, no need to additionally artifical limit it.
If someone thinks that AI content is not better then that’s who the audience is for the remaining human creators
Corpos don’t care, ordinary people don’t care. Does it make it still a good thing that Corpos can pump out slop without paying a living wage to artists or atleast royalties to those they took the training data from (with or without their consent)?
AI can already create better looking photos than I can, but it has zero effect on my desire to do photography. I don’t see what the issue is.
It pretty much can’t. It only mix and pattern matches existing photos.
Coming back to my first half sentence:
AI can’t create and when it only trains on it self it collapes, a short to this:
Gave you an upvote. This is an unpopular opinion shared on unpopularopinion.
Personally I find AI art to look like “AI-art” if that makes sense. It has this generic look and feel to it. It might just be that people have to prompt it differently to get something that does not look generic.
In my view it can be art, but it is just other peoples art regurgitated by a machine with lots of filtering (to prevent nudity and other things that art historically often contain)
I think the biggest issue with LLMs / AI is that it is a loophole to use other peoples work and avoid copyright and licenses.
Personally I find AI art to look like “AI-art” if that makes sense.
I know what you mean and while I generally agree I must still note that this only applies to the “bad” examples. It’s conceivable that you too have seen pieces you truly liked but didn’t realize were made by AI. This is something people often don’t seem to consider; AI art is only bad untill it isn’t.
Art is human creativity. By its very definition, it cannot include something lacking originality.
A person may use AI as a tool to create art, but without human creativity it’s just mechanical regurgitation.
Human is the one with vision, AI is the tool. It’s just a much more advanced paint brush that anyone can use. Alone it doesn’t create anything and if the end result is bad, it’s not the fault of the brush.
AI art is just an art sub-genre like painting, sculpting or photography is. Saying it’s not art is like a film photographer saying digital is not real photography - gatekeeping.
i suppose i cant disagree with the premise… but to clarify, the AI is equivalent to a paint brush or phototshop… a tool used by the prompter to create (extremely derivative and hacky) artworks. i have seen a lot of very expressive works generated by AI, where a concept thought up by the prompter is expressed to humorous or sometimes grim results. but every AI image i have seen has tells of being AI generated.
every AI image i have seen has tells of being AI generated
Except the ones you didn’t realize were made by AI. You by definition can’t know how many have passed for you as “genuine”.
and by the same logic, you cant know if i have or have not been duped by an AI image. thanks for asserting expert knowledge of my perceptionsl capabilities, but you’ll understand that i am extremely skeptical of that assertion. based on how i consume media, the likelihood that i have been exposed to AI generated images without my knowledge is pretty low. but do continue to tell me what my experience of the world is … its kinda hilarious
It’s a rather safe assumption. I too like to tell myself a story about how I can always spot fakes but I know it to not be true.
