Artificial Intelligence & Human Cultures
If you didn’t see someone sharing something they pulled out of DALLE-E-2 the pst few months, your use of social media is likely quite limited. It has to do with creating images from words. Some of these images are funny, others creepy. Then there’s tools like Soundraw that can create music for you. There are others. While they’re interesting and may well be around for a while, does Artificial Intelligence (AI) really play a role in culture? Can AI be a part of human cultures?
For most of us, when we think about culture, we may think of fashion styles, types of music, the language and slang we develop, of art and special festivals and events. This is true. It is also about the social groups we form, reciprocity mechanisms (e.g. money and crypto) and kinship systems (friends, families etc.).
Culture serves a purpose that we often don’t really think about, because for most of us, it just is. We only think much about our own culture when we meet with other cultures, especially ones very different from our own. The real reason humans have culture however, is that it is how we chose to survive hundreds of thousands of years ago. Other animal species chose genetic adaptation, which is very slow. We chose culture, which is much faster. It also got us off the menu of those nasty things that wanted to eat us and we put them on our menu. Food is a part of culture too.
As AI becomes more deeply embedded in our societies, interesting tools like DALL-E-2 pop up and come to our attention. They do so because they are novel and play an important role in helping us figure out their place in our cultures and societies. These types of experiments with technologies are how we’ve decided what tools we want to adopt and which ones we want to toss into the trash heap of technology failures. AI will not be tossed out, but it is in its very early days and still largely experimental.
But while we may see interesting and fun experiments like DALL-E-2 and other tools like music generators, it is important to realize that they’re simply replication and mixing systems, they’re representative, but they aren’t original in the true sense of the word. A tool like DALL-E-2 is recombining, in often fascinating ways, art, images and words that already exist. The same with music, which, at its roots, is mathematical and formulaic in nature.
It is very much arguable that it is art at all when it cannot generate its own art. Art too, is a means of cultural communication. AI does not understand culture. It is not conscious like humans. As culture is fundamentally a survival mechanism, AI does not understand this. I argue then, that it cannot really be considered art in such a context.
We are likely fascinated by the images created from tools such as DALL-E-2 because it plays with, in novel ways, how we have perceived in our lives, what the original images mean to us. We find them discordant because cultural meaning is taken away. The meaning is distorted. We may find it amusing, sad, scary or just weird because it doesn’t really tell us anything.
Most AI music tools today are largely just supportive. They may provide rhythms, but we adapt them, evolve them and go on to create the music we actually want.
DALL-E-2 type experiments are fun. They’re important in the development of Artificial Intelligence in helping us figure out where the boundaries of these technologies rest within our societies and cultures. But for now, as AI is useful only within strict, narrow uses, AI will not be a cultural engine of its own. That, thankfully, remains the bane of being human.
We know AI has racial and gender bias. Work is being done, mostly in academia, not much in industry, to address this. But an entire culture is complex and integrated, far more than any AI tool can be. And an AI tool must be informed by existing data. That is what it is trained in. For now, we haven’t figured out how to fully digitize a culture. That is likely impossible. If you combine data from many cultures, then an AI is generating something entirely new with no reflection on any one culture. But again, we cannot digitize all aspects of what a culture means.
In the meantime, we’re going to see some very interesting and sometimes creepy and weird, AI tools coming out. But we will shape them, not them us.