A few weeks ago, my Twitter timeline got flooded with people talking about their favorite pieces of art, following this tweet. If you’ve been on Twitter during that time, you’ve probably seen it quite a lot of times.
While the hundreds of tweets that I’ve seen were just quoting the post above, the original image in the tweet was possibly the most devastating of all. It’s Keith Haring’s “Unfinished Painting”, which was intentionally left blank by the artist to represent his life being cut short as a result of the AIDS epidemic. It’s a tragic and beautiful rendering of the struggles at the time, and striking commentary on a situation that resulted in the deaths of roughly 40 million people. So, you can imagine everyone’s shock upon seeing this quote-tweet to the original post.
This tweet, to me, is obvious (and good!) satire. Most people took it seriously, and who can blame them now with every other tech bro calling themselves “prompt engineers” because they can roughly describe a scene to a robot. Without getting off-topic, this tweet caused, or rather continued a big argument on the status of AI “art”, and if it really should be considered as such.
Today, the thing we call AI is not real AI. There is no “intelligence”, and I’m not saying that to be snarky because for it to be considered actual “artificial intelligence”, it needs sentience. We merely have tools that utilize machine learning, without actually touching on the so-called intelligence. All the chatbots do is to gather information, analyze human speech patterns, and give you a result that thinks is fitting as a response to what you said. There is no actual intelligence involved, the robot is just solving a big puzzle. And no, I will not get into the fact that “because we call it AI, it is AI” thing, I may be a descriptivist but most people aren’t, and it’d be anti-descriptivist to assume otherwise — but this is a topic for another time.
Having established that the AI in “AI art” isn’t AI, let’s define the “art” part as not art. Without relying too much on definitions since they can change quite easily, especially for a concept as abstract as art, I’ll describe the main, unchanging quality of “art” as something that is a product of imagination. Imagination, that to our knowledge is only seen in intelligent life. Something that our “AI” doesn’t have. People may argue that the imagination part comes from the prompt-writing stage, but that doesn’t directly render itself in the image that is created. Above all, whatever is made using machine learning tools utilizes (and entirely relies on) already existing pieces of human-made media. Even if you were to argue that AI art is art, you’d find that it’s not even AI-made, just like how your essay you copy and pasted from Wikipedia isn’t your essay. Having said that, I am entirely for someone who built the neural net they’re using, and fed their own art to it, to be credited as the artist of the product. But no other case is applicable.