Aight, here’s the thing.

All art is, at its base, about translating a person’s inner concept into an external form. Sculpture, painting, poetry, dance, whatever.

To do any art form, there is a barrier to entry. If you want to be a dancer, some part of your body must be mobile, right? Even if it’s just your eyeballs, dance by definition is about the human body moving.

But, what if you can’t move your body? Is that, and should that be, a barrier? Why can’t a person get an exoskeleton device that they can then program to either dance for them, or to respond to their thoughts so they can dance via the gear? Well, in that case the technology isn’t here yet, but pretend it was.

Obviously, it wouldn’t be the same as someone that’s trained and dedicated to dancing, but is it lesser? It still fulfills the self expression via movement.

That can be applied to damn near every form of art. I can’t actually think of any that it doesn’t apply to at least in part.

There is a difference between a human sitting down (or lying or standing) to write a book and just telling a computer to generate a book. But it doesn’t completely invalidate using a computer to generate fictional text. The key in that form is the degree of input and the effort involved. A writer asking an llm for a paragraph about a kid walking down the street when they’re blocked isn’t the same thing as telling it to write the entire book. There’s degrees of use that are valid tools that don’t remove the human aspect of the art form.

Take it to visual arts. A person can see things in their head that they may never develop the skill to see executed. They may not be physically capable of moving a brush on canvas, or pen on paper. A painter of incredible skill may be an utter dunce at sculpture, but still have vision and concepts worth being created.

The use of a generative model as a tool is not inherently bad. It’s no worse than setting up software to 3d print a sculpture.

The problem comes in when the ai itself is made by, and operated for the benefit of corporate entities, and/or when attribution isn’t built in. Attribution matters; a painting made by Monet is different from a painting that looks like Monet could have done it, but it was made by southsamurai. If I paint something that looks like a Monet, that’s great! If I paint it and pretend it was made by Monet, that’s bullshit.

A “painting” by a piece of software that’s indelibly attributed as generated that way isn’t a big deal. It comes back to the eye of the beholder in the same way that digital art is when compared to “analog” art via paints and pencils. It only really matters when someone is bullshitting about how they achieved the final results.

Is ai art less impressive? Hell yes, and it’s pretty obvious that it isn’t the same thing as someone honing their craft over years and decades. An image generated by a piece of software with only the input prompts being human generated is not the same as someone building the image with their hands via paint/touchpad/mouse/whatever.

This is still different from the matter of using ai instead of paying a human to do the work, which is more complicated than people think it is.

But, in terms of an individual having access to tools that allow them to get things inside their head out of their head where it can be seen, it has its place. It just needs to be very clear that that’s the tool used.

And yeah, I know this is c/fuckai, and I’m arguing that ai has its place as a tool of self expression, and that’s not going to be universally satisfying here. But I maintain that the problem with ai art isn’t in the fact that it’s ai art, it’s the framework behind that that makes it a threat to actual humans.

In a world where artists can choose to create art for their own satisfaction without having to worry about eating and having a roof over their heads, ai art would be a lot less of a threat.

It makes it more accessible to the lazy and talentless.