When I was teaching about glitch art, for the sake of organization we talked about 4 different categories. Data moshing or manipulation – just screwing with the data directly. That’s more or less where most of the effects I use fit in: they are just different methods of screwing with the data – sometimes in unintended ways, but just manipulating the data. Data bending is a form of misalignment – it means encoding or decoding or even processing with the wrong kind of software or algorithm. There’s Circuit Bending or hardware failure – which is hard to capture in the wild, but you can now buy or engineer sort of custom analog gear that can simulate or create this kind of effect. In a way this is what a lot of CRT artists are doing with light painting and forcing analog artifacts. There’s also plain old Misregistration which is just corrupted files. That can be interesting, but it’s obviously unpredictable.
After I had gotten a handle on generating imagery with Processing and raw materials and it started to get a little predictable what would result, I wanted to be surprised again. I tried my hand for a while at something I used to teach that fit under the umbrella of Data Bending – using Audacity and audio algorithms on images in their raw data form. The results were unstable, often boring, and underwhelming – but when I managed to churn out a few interesting things I had this thought: it’s like working with aliens. It’s like working with something alien, at any rate – because it doesn’t do what a person would do if you were working with them.
It’s not exactly like AI though, because it really has no idea what it’s doing when it misapplies the audio filters. Interesting side-note, because of the Chinese Room thought experiment on AI, this is the entire reason I don’t really consider GAN art to be AI art. The thought experiment in a nutshell – if a person who did not speak Mandarin were confined to a room, and fed tiles of Mandarin characters through a mail slot and could only get food through the slot if they rearranged the tiles correctly, they could be making good sentences in Mandarin but never know it – because they were just trying to get food. No matter how complicated the Mandarin tile arrangements became, that person can’t read it – they just rearrange symbols until they are told they are successful. So GAN “art” can be pretty or cool, but it’s not really art anymore than the person in the Chinese Room could write literature. This ties back again to my earlier discussion of Gibson – in Mona Lisa Overdrive and Count Zero (spoilers) he deals with the phenomenon of an AI that is an actual artist. Meaning, it’s not just returning information, but understands and is somehow manufacturing semantically significant artwork. I think a lot of people don’t grasp that this is a major leap – this is the revolutionary thing about that AI, that it somehow escaped the Chinese Room dilemma.
So there is a series of which only a few have been posted here and there exploring alien intelligences. The best myth about alien intelligence I’ve come across is the crop circles message encoded in binary that reads:
Beware the bearers of FALSE gifts and their BROKEN PROMISES.
Much PAIN but still time.
BELIEVE. There is GOOD out there.
We oppose DECEPTION.
The closest thing I’ve found that sounds like this kind of chaotic AI that somehow stumbles into something aesthetically good also happens to be this black midi track called Arecibo. You can hardly watch and listen to that and not feel that an alien intelligence or an artificial intelligence can turn around at some point some kind of art that isn’t just surprising, but transcendent.