I think everyone has more or less decided their stance on AI art. It's hard to find someone who's not all for it, dead against it, or apathetic to the entire thing. Myself, I was dead against it, but decided to start using Stable Diffusion so that I could understand how AI art even worked. What I discovered was that AI art is a horribly named tool that has the potential to be great but might be getting misused.
For the uninitiated, AI art generators work by filling an image with a "seed," which is just colored noise/static. It then looks at your prompt and uses the images it was trained on to shape the seed into the image you want. It's really just an art sorting algorithm.
The biggest grey area with AI art seems to be that they've all been trained on scraped images. People for AI art believe that shouldn't matter, as most artists learn by copying things they see. I'd say what AI does is a little different as it has the potential to perfectly replicate these images. Most artists will copy from life, other artists, etc, but will eventually come up with their own style. AI isn't capable of doing something like that. Due to it being an algorithm, it's kind of stuck piecing together images from its library. We could talk about the limitations of AI all day long, but it's been getting better at an exponential rate. It can generate beautiful faces, and will eventually get hands sorted out too. But it will never be able to create something truly original. Whatever it makes will always be an amalgamation of the images it was trained on.
And that's the one hurdle that AI art will never be able to jump. It needs tons of reference images for something to draw it accurately. By its very nature, AI art cannot make anything truly original. You need thousands of references for the AI to draw on before it can replicate it.
I don't have a strong opinion on "AI art artists" getting paid. It takes work to make an AI image look good. The art might be created in seconds, but then you need to regenerate the bad looking parts, or port it over to your favorite paint program for a touch up. I laugh when people submit art they made to art station and are too lazy to remove the extra leg their character was generated with. The AI may be stealing artwork, but there are a lot of "starving artists" that have no respect for copyright laws and have made careers out of drawing IP they don't own. Art is one of those things that how much it's worth is very much up to the person who is buying it. especially in the case of commissions.
I can see why less known artists are scared of AI. Especially places like Pixiv where most of the artists have made a living out of drawing "character X" in the only ten poses that ever get on the front page. Places like this where originality is dead are ripe for an AI takeover. Perhaps AI is the great equalizer we need to push artists away from tried and true methods for success and to start taking more chances with their art. Or maybe we'll sink deeper into a sea of unoriginal works and styles as everything becomes a single bland amalgamation of everything. Just like what the you tube algorithm did to most of its creators.
Also AI written works are hilarious.
For the uninitiated, AI art generators work by filling an image with a "seed," which is just colored noise/static. It then looks at your prompt and uses the images it was trained on to shape the seed into the image you want. It's really just an art sorting algorithm.
The biggest grey area with AI art seems to be that they've all been trained on scraped images. People for AI art believe that shouldn't matter, as most artists learn by copying things they see. I'd say what AI does is a little different as it has the potential to perfectly replicate these images. Most artists will copy from life, other artists, etc, but will eventually come up with their own style. AI isn't capable of doing something like that. Due to it being an algorithm, it's kind of stuck piecing together images from its library. We could talk about the limitations of AI all day long, but it's been getting better at an exponential rate. It can generate beautiful faces, and will eventually get hands sorted out too. But it will never be able to create something truly original. Whatever it makes will always be an amalgamation of the images it was trained on.
And that's the one hurdle that AI art will never be able to jump. It needs tons of reference images for something to draw it accurately. By its very nature, AI art cannot make anything truly original. You need thousands of references for the AI to draw on before it can replicate it.
I don't have a strong opinion on "AI art artists" getting paid. It takes work to make an AI image look good. The art might be created in seconds, but then you need to regenerate the bad looking parts, or port it over to your favorite paint program for a touch up. I laugh when people submit art they made to art station and are too lazy to remove the extra leg their character was generated with. The AI may be stealing artwork, but there are a lot of "starving artists" that have no respect for copyright laws and have made careers out of drawing IP they don't own. Art is one of those things that how much it's worth is very much up to the person who is buying it. especially in the case of commissions.
I can see why less known artists are scared of AI. Especially places like Pixiv where most of the artists have made a living out of drawing "character X" in the only ten poses that ever get on the front page. Places like this where originality is dead are ripe for an AI takeover. Perhaps AI is the great equalizer we need to push artists away from tried and true methods for success and to start taking more chances with their art. Or maybe we'll sink deeper into a sea of unoriginal works and styles as everything becomes a single bland amalgamation of everything. Just like what the you tube algorithm did to most of its creators.
Also AI written works are hilarious.