I think one of the problems of how we assign value (or lack of value) to AI-generated art is that, often in these debates, the origin of said art is known. It’s easy to browse a gallery of AI-generated pieces and tell yourself, “Ah, these do not evoke/stir any feelings in me” or “Look at how soulless these are,” but the issue is that confirmation bias can subconsciously kick in. I think a more objective study examining the value or enjoyment of art by having users rate which of two pieces they think is more artistic or that they like better with a large sample pool of human-derived art with AI-generated art mixed in would be a really interesting study to see the results of. I have a feeling these studies might already be underway or maybe there’s already some website that has set this up as method of “testing” your proclivity towards human or AI art. Note that these studies would be somewhat hard to set up in a rigorous manner–for example, I would think you would have to stick with one type of artistic style (to rule out subjective favoring of certain styles) and you’d likely have to try and populate your human-generated art with pieces that aren’t well known, to remove other biases.
An interesting test (that probably won’t happen) is for the Pokemon company to quietly release AI-generated illustrations under a pseudonym, trying as best as possible to stay with a consistent style, to give the illusion of a human behind the works. It would then be interesting to see how the reaction to said “artist’s” illustrations would be, and if AI-generated art truly can somewhat “replace” human-generated art. Or an even bigger question behind an experiment like this: if you don’t actually know the true origin of said art (you have no way to confirm if a particular piece was definitely drawn by a person), does it really matter? If we can’t confirm the origin (and as consumers, we can only somewhat know because of the credit the artist receives on their card), do we then just objectively evaluate these illustrations based on pure aesthetic value? If this is the case, how would AI-generated art measure up against human-generated art then? I think these scenarios could already potentially be playing out before us:
With an assumption of a human behind the works of AI-generated art, the lines become even more blurred and really challenge what we think about AI art.
This is why I said, AI is not the artist, but the tool. If we consider AI art from that PoV, it’s much easier to evaluate the merits of it, whatever it is. As a performer, I keep coming back to improvisational art. There really is no truly original improv art, but that’s not the point of it. The spontaneity is. That aside, I copy others, and they’ve copied me, and honestly, I can’t think of anything I’ve ever done that was not inspired by something done previously, or inspired by nature and the organically random world around me. (ie. Hearing a sound pattern somewhere, like a construction site, and that inspiring a rhythm I develop later.) I still say, we have to look at it as “AI is a tool for art, like a paintbrush, but it doesn’t solely create the art.”
If we just viewed AI as a tool–as the “paintbrush” of the prompt generator–and assess the pieces created independently of the controversy behind their generation (i.e. distilling it down to “Can AI generate good art?” like you mentioned previously), I think most would probably admit that AI can certainly create technically good art (obviously there are still some kinks, like with hands and things). Technically, AI-generated art won’t break new ground or create new styles unto themselves, but it seems fully capable of being “inspired” by similar pieces of art and replicating them well. And, like you mentioned, inspiration is everywhere these days–true originality is far and few between. Most artists do not just come up out of the blue and invent their own styles–they copy others, tweak it, and find what works for them. In this sense, then, I think AI can make “good art.” (Of course, the term is relatively loaded, as “good art” is completely subjective).
But I think relegating AI to just being the tool of an “artist” is simplifying the heart of the debate. You brought up the question of, well, what of the prompt generator? Is he then considered the artist of the generated pieces? I think many would still take issue with viewing any and all prompt generators as artists. For those that have trained in art or were artists before AI rolled around, I would think many would not have a problem if they incorporated some AI in their works, either as soundboards to bounce ideas off of or even in possibly just presenting purely AI-generated images. But what about those that know how to maneuver the prompts but that have no real artistic bent or training? I think one of the main issues of the debate over AI art lies here. I don’t know if most people are comfortable with, all of a sudden, anyone being able to generate beautiful imagery and acquiring the title of “artist” for being able to type descriptive sentences. It reduces years of experimenting, learning, and practice into the simple ability of typing. I think this is one of the areas in which the “soul” of art is crushed.
Good thoughts, But…
I’ve worked with a few major (and many, many minor) professional choreographers and performers who will staunchly defend the stance that ANY and ALL people who seek to move their body are artists within dance, even if they are not trained. Skill is just what allows your voice to be more heard, but we are ALL artists, as innately creative beings, they would say.
Part of being an artist and being successful as one, seems to require that you express with your own more unique voice. This may be the distinction that AI as a tool cannot give, and artists who use AI cannot achieve. Anyone using AI/chatbot will create similar work to anyone using that same chatbot. I’m not sure here. Just a thought.
You have a point there. Depending on how you define an “artist,” you can reach substantially different conclusions on whether you would or would not classify the prompt generator as one.
For the second point, I think you could reasonably still find your own style or voice with AI-generated art. For example, I could see someone on Instagram or whatever be known as “the person that depicts Star Wars characters as cats” or something. And that may be what that person likes doing and what he/she becomes known for, all through AI-generated art. Even if someone else could prompt the AI to make similar-looking illustrations, the “Star Wars cat person” may differentiate themselves through their pure interest and sheer dedication–e.g. generating at least one unique image every day for more than a year.
All interesting thoughts though, and the discourse has been fun to think through
@bk2021 This is certainly a valid perspective that other people share. I don’t quite view AI-generated art to this extreme (i.e. stealing from artists). I view it more as condensing years upon years of experience into a short amount of time, as I think many people who evolved as artists did so on the backs of other artists, through copying/studying other peoples’ works and learning from them as a result. Imitation is normal in art, and AI just takes this imitation and learning to an (uncomfortably?) extreme level.
When I worked with AI, machine learning was always described as far less complex than human or actual learning. As an example, non-guided learning tended to produce a breakdown of usable patterns, unless, theoretically, the size of the neural net is HUGE. Like HUGE HUGE. At this time, we can’t do that, so I wonder if a more nuanced AI could more practically filter and compress learning time frames, thus producing art that is more acceptable, yet still considerably unique.
Generally, I’d agree with you that someone who asks a chatbot to make art for them is not an artist. BUT, I have bad news for you: ANY, no, EVERY performing artist in improv - comedy, music, dance, acting - steals from others. It’s just the nuance that we use when we borrow, riff-off, or “sample”. It’s a time honored practice. That IS how it works. SO, as a professional performer and teacher, I’m offended and saddened you think we’re not actually artists. Kidding. I’m not offended. BUT, do consider this fact of practice.
On the one hand, I am appalled by the insane copyright length - something like 75 years after the death of the author - which renders 99% of works irrelevant or lost, but on the other having no copyright whatsoever is just calling for exploitation.
I think my main problem though is that the abusers are always going to be the same - corporations, people with more money/better lawyers etc.
Will AI be for the common good or will it be paywalled etc.
Also @xileets I’d argue for the artist the main concern/purpose is not monetary/exploitation which is undoubtedly the main concern of those developing/owning these AI programs/chatbots etc.
I don’t trust “tech bros” at all since they’ve shown themselves time and again to be selfish, fascist and egotistical with god complexes.
It’s still the same. Nothing has changed from the “hey computer here is an equation and some data, can you optimize the best parameters?” We are still just line fitting at a huge scale and there is zero intelligence.
On the surface it might appear that AI can do what artists can, but it really can’t, and I doubt that it ever will be able to. All AI can do is recreate what it understands as patterns with the data it’s been fed. It’s doesn’t understand what it’s doing, and that’s the huge difference here and what I think people should keep in mind. When an artist creates art for a card, they are making choices based on the story they want to tell through the visual medium. The colors, the composition, the subjects… they are all done deliberately to achieve a specific end goal. The AI doesn’t understand how certain colors or compositions make people feel. All they know is “this is something that a lot of other artists do, so I’ll just do that too.”
Maybe this sounds a bit too pretentious. Sure, AI art will be indistinguishable to a lot of people, but it’ll likely never be able to actually recreate the artistic process.
100% agree. Actually, I was arguing chatbots in analysis with a peer the other day, and there were far more pretentious stances. haha (Artist turned IT pro vs. IT pro)
I will say, AI as a true AI in the future will be able to create art, but chatbots, or “AI” that we have today, is not actually even close to AI.
Good point on the art-for-money thing. As an artist, Ioften had problems becuase my art was driven by my love, not the money…
For the “bro” attitud, I’d argue that any “___ Bro” is going to be,
Though I don’t know about fascist… yeah. probly. haha.
Techies run the gambit from idealist-hopefuls to abusive manipulators.
If the inventors of the internet would have been more authoritarian, we’d have fewer problems with security and cybercrime. But there are benefits, like this fourum.
Kidding. I’m not offended. BUT, do consider this fact of practice.