More thoughts on AI art discourse - there’s a claim I’ve seen pop up a few times that AI art models are tools for “democratizing creative expression’ because they enable people who can’t draw or paint well to make beautiful art about things they want to see. This claim is weird to me, in a way that’s had it rolling around in my head since I heard it. I have zero incentive to begrudge the use of technological tools to help create art - even those that embed strong priors in what that art output looks like that comes more from the technology than the user. That statement applies to many interesting and well-respected mediums, including those that act like AI art in other ways - remixing and resampling existing work is a legitimate way to express creativity. But like, you don’t make a collage of something and tell someone ‘I drew this’, right? Or you don’t say, “this is just like if I drew this”. Which is what I feel like this claim is saying. That making AI art of something will be as if you drew it yourself. 

That’s a claim that will elicit immediate backlash in the same way that traditional art theft (“I made this”) would. But I think it’s different than claiming the skills and work of another artist directly. Most people I’ve seen sharing AI art are honest about how the image was generated. The benefit they get isn’t the clout of having the skills to make the image - it’s just the clout of the image… existing, I guess? So I don’t think that’s my issue with the framing. I think, instead, the part of the original claim that stirs me up is this idea that people who use AI models to make art would not otherwise be able to creatively express themselves on as “high a level” without AI tools. And that’s weird to me! 

I think it’s common to think of the process of creating art as of translating some platonic ideal of what that piece could or should be from the artist’s brain into an actual artifact. The artist wants something to exist, and they bring it into existence, in a form that’s impacted by not just their self-expression and decisions, but also their physical abilities and the tools available to them. It’s not uncommon to metaphorically label these two parts of the process into something akin to “signal” and “noise” - to imagine that the ideal art in someone’s brain is corrupted by their lack of ability to realize it. In my experience, most people who don’t think of themselves as artists think that this is what making art is. You get better at art to reduce the ‘noise’ and make images that look more like you want to make. 

I’m not gonna claim that’s wholly untrue. I think it would be dishonest of me - god knows I prefer digital drawing over traditional because it gives me tools for making my art cleaner, clearer, and more deliberate. But also, the more art I’ve made, the more I’ve come to reject the “signal/noise” framing. Artists are not printers. It is much more common, in my experience, to ‘think on the page’. When you’re making art, you’re communicating with the world, your medium, your own abilities, at every step of the process. I don’t see those influents as noise - or rather, when and whether they are benefitting or hindering the art is extremely contextual to what you’re trying to do. The “democratization through AI art” claim comes in with some very specific ideas of what all artists are trying to do. It implicitly assumes the goal of art is to produce a very polished image in a specific and recognizable style, that features a desired subject or theme, and is appealing to look at. It then follows that people who have creative ideas but who aren’t able, currently, to bring them to fruition in that exact way, are deprived somehow of the experience of making art. 

I disagree with that. I recognize that technical art skills and expressive capacity often grow together, but art that is made by a person without those technical skills is just as art-y as what they’d make with years of training. The quality of expression comes from the process of making the art, not the polish of the outcome. I know there’s an eternal, unresolvable discourse about how much certain tools erode the artistic process. Again, the use of new technology to create art in general doesn’t bug me. What does is equating what I see as processes on very different parts of a spectrum - saying they’re doing the same thing, and fulfill the same role for an artist, audience, or art as a whole. (Plus, there’s a whole other piece to be said about seeing the way that art is shared and distributed online currently, where art with certain polished visual styles is more likely to be produced and shared en-masse, and then thinking that the way to make that process more ‘democratic’ is to give everyone the ability to make art that looks exactly like that, rather than to consider and elevate the art that people are already making.)

Which brings me to my last point- I don’t actually think it’s bad in any way to want to have a polished piece in a specific style that you didn’t have to draw yourself. But that need isn’t met by better drawing tools - it’s met by art commissions. The claim that AI models make it easier to make art, rather than to bypass paying artists to make you art, is a spin that should have more eyebrows hitting the ceiling.