Artificial Ethics
It's mostly silly to be discussing the moral implications of AI generated visual art when AI generated literature has been going on beneath everyone's noses for years
The main problem with all of the new discourse on AI-generated art is it's focused only on the visual arts. People are buzzing about AI art; the topic seemed to have popped up out of nowhere and exploded over night. I guess "out of nowhere" is the wrong characterization, though, because the reason it's on everyone's minds is DALL-E, DALL-E 2, and Google's newly released AI Art Machine. As these generative tools are time-released to the general public, the internet is flooded with the generated images, screenshots of the options, and screenshots of the prompts people are giving to the Intelligence. It's a whole new treasure trove of things to critique and thinkpiece on. Content begets content. Art and culture writers are gifted more useless shit to commentate on. Blessed.
It's particularly hilarious to hear the contra-takes against AI art coming out. I love hearing people say that AI art is demonic. The logic behind this argument wasn't immediately apparent to me because the way I see it, an AI artwork is only as evil as the intention behind the prompt that's fed to it. I'm quick to put all the blame on the person who's deciding what bidding to give to the generator, but that's just my reflexive instinct. Putting more thought into the situation, I'm beginning to understand what is making people afraid. If we understand art/artists to be divinely bestowed with the gift of aestheticizng the world, then giving up that power to a GPT neural net, and anything it produces, does seem perverse. All other arguments against AI art derive from this main thesis. Whether one believes in "God" or not, AI art is still an issue of spiritual integrity. What makes an artwork "satanic" in particular? First of all, we know that art can be satanic because we know art is spiritual, but second of all, we know art can be satanic simply because we are judging certain artworks to be so. When humans have an inkling, or notion about the spiritual value of something, that tends to mean there is inherent spirituality in it. One argument that’s been a refrain for the past several hundred years is that morally/spiritually “good” art must be beautiful to be good, since beauty equals truth and truth is the ultimate good etc etc. So, given this, since most generated art is ugly, then it cannot be good. This line of thinking is especially abundant in the NFT discourse, or has been, at least until Cultural Trend Czar Dean Kissick pointed out that the tastelessness is the point. In the latest "Downward Spiral" there's an understanding that NFT artists understand what they're making is aesthetically bad and they approach this practice with earnestness. I guess this means, then, that there's an enduring truth to NFT's... which endows them with an inherent goodness.
I think what makes art satanic or demonic is the mimicry, which is what most people recognize in AI art and why they instinctually understand it to be bad. However, mimicry is not a hallmark of GPT software. Several American Modernists were already dabbling in demonic mimicry decades before all this during the CIA Abstract Expressionism psy-op. Whether the artists themselves knew it or not that their experiments in Modernism were weaponized in a cold culture war, they did understand the fundamental truth that art is spiritually potent. All art is religious, I say, and the expressionists had been trained in aesthetics. They knew the symbolic attachment that certain colors had in medieval and renaissance religious art. They learned the psychological effects and impacts artworks have on the general public. They did not want to create explicitly religious work—obviously, they didn't even want their art to be explicitly representational—but they did want to manipulate and hold power over their viewers, so they mimicked what religious art had been doing for centuries. To no one's surprise, it worked. Mark Rothko even has a "chapel" for his work.
But visual AI art isn't even new or cutting edge. Feeding prompts into a GPT2/3 machine learning matrix has been around for ages, it's just that being made widely available and the internet proliferation of it all makes it appear contemporary and cutting edge. Maybe this is why I find the commentary sort coming out sort of boring and tired. People are reflecting to themselves, "What are we to make of images created by robots?" or, "What should we be anticipating now that there are AI's making paintings?" AI images generated from text prompts are not the only kind of AI art that's out there. What's impressive and new is that neural nets are able to take text and generate images from it, which is pretty interesting and innovative. But the programmers who are focused on refining the training schema of neural nets for natural language processing are making greater strides, and the literary arts are what we should be paying attention to. We can expect that DALL-E 's or Art Machine's images are going to be more aesthetically bland than interesting or beautiful, but what about the AI's that have a writing practice?
Tivon Rice is a New Media artist who works across disciplines like photogrammetry, virtual reality, and artificial intelligence. For one of his works Models for Environmental Literacy he trained three different neural nets on different datasets—texts such as novels, essays, poetry, and formal climate change reports—and proceeded to co-write a script for his montage film with the AI's. In the video, viewers are listening to computer voices narrate this script. Each narration has a different prose style, taking after whatever textual dataset it was trained on. Rice "co-wrote" the scripts in the sense that he edited out weird semantic artifacts, repetitive words, or phrases that made the narration nonsensical. What was kept in tact was the style that the neural nets had adopted.
I think it's mostly silly to be discussing the moral implications of AI generated visual art when AI generated literature has been going on beneath everyone's noses for years. It makes sense, though, because the internet paradigm we exist within is largely dominated by images. Sending images back and forth is easy and instantaneous. More than this, though, we are visual creatures and this is the reason why the visuals have dominated human imagination for longer than civilization. So we care more about what the robots are painting because we view that as an indicator of how threatening the singularity really is. We should turn our attention to what the robots are writing, though, because that's what is going to be shaping our minds, and maybe even without our knowledge of it. The customer service chatbots were one thing and the Turing Test was maybe our last hope of a guardrail against letting the bots get the best of us, but the NLP models are only getting better attuned.
Artificial Ethics
Another excellent piece with eloquent writing, and it made me wonder if someone should write a short story about God having an identity crisis when humans start inventing and creating things on their own. Or does God take joy in seeing humanity create? Is the commandment against “graven images” something that needs to be taken into account on this point? Might have to think on that…Anyway, always great work Bethany!
"We should turn our attention to what the robots are writing, though, because that's what is going to be shaping our minds, and maybe even without our knowledge of it".
One idea I've seen is AI could be a good way to soft block people. So say you have a toxic person in your replies. Say you know they'll create a new account just to harass you. That's where AI comes in. You train an AI on your posts and every time they reply to you, they get a response written by an AI. What would be even better is if the AI was trained to de-escalate and sooth the toxic person.
As to why we care more about visual art: about a third of the brain is dedicated to visual processing. We can take in an image all at once.