Artificial intelligence, or AI, is transforming art forms that have existed for millennia, as have other technology before it. It will also be useful in the development of hitherto unimagined art forms. In the future, there will be three or four types of intelligences creating art: human artists, humans and AI working collaboratively, machine learning models that do not require human initiation but still lack sentience, and finally, if technologists create truly sentient artificial intelligence, AGI capable of self-creating art. In all likelihood, AI will not kill art, but it will modify it—a process that has already started. Humans are making decisions about how that will happen, and they are doing it right now.
In the late 1800s, the camera did not replace painting, but it did change the way visual art was made. Artists experimented with aesthetics and meaning, free of the compulsion to reflect reality convincingly. At the beginning of the 20th century, the readymade (the use of found and manufactured things as art) did not kill sculpture, but it did free some artists from ideas about originality and craftsmanship that limited their work. This allowed sculptors to concentrate on theory and ideas.
As a result, one might argue that the world has always raised existential concerns about new technology. In fact, comparing machine learning to a 19th-century camera that cannot make decisions or learn—much less run itself—is not entirely accurate. As we move toward an AGI future, combining human practices with machine learning brings us closer to a clear horizon where human and machine intelligences overlap and work together. Or is it a cliff?
But it’s also worth considering how people will be utilized in return. Humans create the models and, in general, give the training data. It’s already clear that just scanning the internet for training data includes blatant prejudice that puts underrepresented populations at risk. Today, rather of turning a blind eye and hoping for the best, it is critical to construct the Ais training process with aim.
Furthermore, the inputs for existing machine learning regimes originate from the same artists who may be replaced by the models’ paid work: illustrators, designers, authors, musicians, and translators.
But if artists and people who support them want systems that build culture instead of destroying it, artificial intelligence (AI) could be a great way to learn and work together. That is a huge if. The mechanisms for teaching AIs are far from complete, and people developing AI algorithms are making key choices right now.
If creative growth is to continue, now is the time to put money into supporting and growing the arts, education, libraries, and culture in a fair way, instead of taking the easy way out and letting a few corporations take as much as they can from existing artistic treasures now, hoping to sell them back to advertisers later. Humans have built a society with a lot of wealth and power, but not everyone has access to it. Technological and artistic innovators and visionaries can design the world they want. People, not AI, make that decision.