Statement on AI Art

It would seem many artists are putting out statements with regard to AI art. I don’t know if I’m statistically significant enough for my statement to mean much, but in the interest of complete transparency, I’ll share my view anyway.

I am firmly of the belief that AI art is here to save us, not destroy us.
Allow me to explain.

There was a time when individuals needed a great deal of technical skill at programming in order to make a videogame. In the days before software like GameMaker Studio, Unity, RPG Maker, and other game development engines, videogames were programmed the “old-fashioned” way. When game development engines came out, many disgruntled programmers were upset. They complained that the new software would make it so that anyone can make a videogame, and that many of them will be terrible. And, to an extent, they were right.

I won’t lie, the videogame market is oversaturated with trash games. And yet, the indie game genre may never have gotten off the ground in the first place without game development engines. Many sensational game franchises have started out as indie productions: Minecraft*, Undertale, and Five Nights at Freddy’s, to name a few. These games became popular enough to rival studio-produced games. That is significant. The advent of game development engines shifted where the value of videogames lies. Originally, it was merely a test of your programming skills. Now, it is largely a measure of how good the idea behind the project is.

By contrast, independent films are rarely as competitive. A few manage to gain some renown, but the household names of film are always owned by big companies: Disney, Marvel, Paramount, etcetera. Why? Money and manpower. It is easier to make a high quality animated film if you have the budget to hire a whole team for the process. When it comes to the film industry, big companies almost always have the upper hand over independent creators.

So, what’s this got to do with AI? In my personal opinion, AI has the potential to level the playing field. There’s been a lot of stir recently (this is being written on 23 Feb 2024) with regard to OpenAI’s SORA, which produces remarkably consistent and coherent video clips. I’ve heard people say that it’s going to kill the animation industry. I disagree. I believe that this is the equivalent of game development engines. This technology is improving at an exponential rate, and the value of art and animation is shifting. It is now no longer merely a test of your ability to “just pick up a pencil”. It is now, like with videogames, a measure of how good the idea behind your project is.

This will produce a lot of low-effort AI-generated slop. I think everyone in the art community has seen the endless slews of generically-attractive anime girls produced from this. Likewise, there are countless low-effort indie games. And yes, the oversaturation is an issue! But I would say that the gems we’ve gotten from the indie boom is well worth the slop. I believe that the AI generated films we’re about to see will be largely trash, but that the ones with good ideas behind them will really be able to shine.

It’s probably also worth noting that there are a lot of popular artists - “real” artists - who only really draw generically-pretty women with no real story behind them. I won’t name any names, but you’ve probably seen their art around. These are the artists who are threatened by AI. Because anybody can generate an image of “Pretty girl #5280”, they have nothing left to bring to the table. The artists who will survive AI are the ones with impactful stories behind their work.

Now, you may notice that I have not addressed the common assertion that AI art is theft. I have avoided this because, truthfully, it is in a grey area. I believe that artists should be able to opt their images out of AI training for the simple reason that, if I were inspired by another artist’s work, and they asked me not to make anything based on their piece, I would respect that wish. But, if I went ahead and did it anyway, as long as my product was sufficiently transformative, I would not be violating any copyright in any meaningful legal way. I don’t believe that AI training meets any legal definition of theft, but I also think that artists should still be able to opt out. To this end, when I do use AI in my own work, I never prompt with the names of other artists. I use a model trained on copyright-free data, combined with a textual inversion embedding that I created based on my own art style.

Finally, when discussing this with a fellow artist, they brought up the valid point that AI text generation is improving every day as well, and there may well be AI generated films based on technically-good but AI-generated stories and ideas. There is every possibility that this will happen. In fact, I am almost certain this will happen. But it doesn’t worry me either.

Disney’s The Lion King is an adaptation of Hamlet. It did not displace Hamlet from its position as a well-known piece of literature. This fact also does not mean that we, as a society, did not “need” the Lion King because we already have Hamlet. It would be disingenuous to pretend that the Lion King did not bring anything new to the table in its adaptation. If AI beats me to my own ideas, that will not mean that my stories are not needed, as they will bring something new to the table. If my stories are later used to inspire an AI production, that production will also bring something new to the table, and whether that “something new” is a human element or one generated by AI, I really don’t care. I will support it regardless.

* It is worth noting, of course, that Minecraft was coded the “old-fashioned way” in Java and not made with a game development engine.
← Back to Home