Is AI Art Real or Not?

To illustrate my points in this article, I asked AI to generate “real art”. Created with Midjourney.

Exploring the question with a point-counterpoint

Artists beware, generative AI is here to steal your jobs! Or maybe not. If you’re an artist yourself, are you worried that this new technology might replace you? Whether this expectation proves true or false in due time, I’d argue the big question is whether generative AI can do real art at all, whatever “real art” may be.

Do I have a definitive answer? Of course not. But that’s not going to stop me from having some thoughts. Let’s explore some of the possible answers using the format of point-counterpoint, or a mini-debate between me and myself. I’ll give my best attempt at steelmanning both sides, and if you have any thoughts of your own, feel free to continue the debate.

Considering which opinion seems to be more common online, or at least louder, let’s make the PRO side the claim that generative AI art IS NOT real art. The CON side will therefore be defending the idea that generative AI art IS real art. Ready, set, go!

This is probably AI thinking to itself “Oh god, what have I done?” Created with Midjourney.

PRO #1 – Generating AI art takes no skill or effort.

While it is true that many “AI artists” today are having their “Skrillex pressing spacebar” moment in the sun, pressing a couple of buttons truly doesn’t take an artist. The difference in the amount of skill and work it takes to produce a piece of art the good old fashioned way versus using generative AI is staggering. It’s years of training, trial and error, and hard work.

To the extent to which the output of generative AI art can resemble real art, there was real skill and effort involved, but not on the side of the art generator. The AIs were trained on real works of real artists, usually without the real artists being compensated, only to allow anyone to mimic their style in seconds. If this technology takes away the livelihood of real artists, that’s it. There won’t be much more new real art produced in the future, including for AIs to steal.

CON #1 – It’s just the “photos aren’t real art” and “games aren’t real art” panic all over again.

Where have we heard this type of argument before? It feels very familiar. Right, whenever there’s a new technology that allows for a new form of artistic process or expression, the old guard declares it to not be real art. Until, inevitably, the new generation universally agrees about two decades later that the new artform of course is real art, just a new kind of it.

As for the skill and effort argument, while it does have some merit, it’s a bit like when Red Letter Media was making fun of the movie Boyhood. Apparently, the movie must be good because “It took twelve years to make!” Art doesn’t have to be hard to make to be real. Ask any graphic, literary, or musical genius. Also, how hard is it to take a photo? Don’t you “just press a button”? If prompt engineering is the new artform, it will see its own effortless Mozarts and toiling Salieris.

Here, the monster of AI is about to devour the child of creativity. Created with Midjourney.

Pro #2 – AI art isn’t creative.

Even if prompt engineering was a real artform somehow, all of its possible results can only be drawing from a pre-existing pool of art that was created in the past. The AI may remix and reshuffle, but it doesn’t create anything new. It takes human creativity to produce something new, and without that, you can’t have new real art.

Con #2 – Technology is neutral, it matters how people use it.

It’s true that the generative process of AI is more like exploring a pre-existing space and finding cool things in it, but again, how is that different from photography, for example? Just like you hike through the real world to take a photo from a new point of view, you ask the AI in a new, unique way to show you some part of the artscape that nobody has ever looked at before.

Or you can ask the AI for a big-breasted anime waifu in the style of Pokémon, or take a photo of a cheap copy of Mona Lisa. The artist’s choices make it art, or not.

AI rendition of real art dying alone in a forest. Created by Midjourney.

Pro #3 – Art requires humanity.

For a work of art to be real art, perhaps it ultimately doesn’t matter how good or new it is objectively, it matters that it was made by a real human person.

Con #3 – Art is what people decide art is.

Or, perhaps, it matters most what most people believe. If people decide to only buy art from human creators in the future, then real art is what humans make. But if in a generation or two, everyone is okay with AI-generated art, then it will be real art.

What do you think? Do you agree with any of these points? Do you vehemently disagree? Did I forget anything? Let me know.