Generative AI and Definitions of Art

Feb 7, 2023

          When Google released their MusicLM paper, revealing a sophisticated prompt-based text-to-music audio generation model, it felt similar to Dall-E or ChatGPT: a tool that promises to revolutionize what it means to create, promises to democratize art through artificial intelligence, promises to open up the infinite possibilities of human creativity (or something along those lines).

          I think one of the biggest perils of this kind of technology is the promise of the removal of friction. Almost every kind of daily technology revolution comes with some kind of image that your life will be smoother now, that downloading this app or installing this extension or using this tool and buying this product will take away all the pesky work and let you streamline your life so you can open up time and space for what matters. Perhaps that’s true, we can now create full paintings, songs, and stories in a minute that arguably rival the quality of work of lifelong servants of the craft.

          A Substack article by Rob Horning reads: “Much as automated customer service makes it seem like a luxury to get human assistance, the ubiquity of automated content will gradually make the demand for thoughtful human-made content seem more extravagant.” And it’s true, as every experience around us is transplanted to a screen and streamlined with a minimalist white UI in some technologist’s frictionless bland wet dream, people are still finding ways to reintroduce friction and time, because time and effort is what makes life meaningful.

          People buy disposable film cameras to force delayed gratification, they spend money on handmade mugs and leather wallets that are special because they take time and effort to create, they set up a whole coffee or cocktail bar in their house and master the craft even though they could buy canned versions at Safeway, they walk to class instead of driving because they like to feel the sun and hear the birds, they spend hours knitting a sweater from scratch even though there’s a $3 version on Alibaba, they drive to see their friends and family in person because texting isn’t enough, they paint and sing and write because they know that an AI won’t compare.

          As I consider all of these new AI tools and whether or not they are “art” (at least within my own personal definition), it’s those examples that stick out to me. Art, to me, is anything that requires thoughtfulness, intentionality, and often time. To me, what makes a piece especially striking or beautiful or inspiring or engaging is the thought that another human being chose to put in the effort to express some sort of aspect of their life, to share their unique perspective with the world through their medium of choice.

Furthermore, it’s that unique human perspective that is necessary to create revolutionary art, to truly internalize the world and find a new way to  interpret the human experience. In that same article, Rob Horning also writes that “Our interactions with [AI] further reduce language and images to a set of statistical relationships that can be infinitely recombined without ever producing anything new or potentially destabilizing."

          While there’s something to be said for lowering the barrier of entry to art, music, and writing, I think that lowering that barrier takes away that special element of thoughtfulness. In fact, the selling point of AI is often the thoughtlessness: there is no time delay to force you to consider more deeply what exactly you’re creating and why. While this kind of thoughtlessness threatens the definition of art, I don’t believe that my definition will ever go away. Despite all the hype around the new generative systems, I consider them simply toys, and I believe that people will continue to value signs of humanity, to value the hand of the artist, and to praise the effort that goes into truly revolutionary or beautiful artworks.