AI Musings I: Art
“Once men turned their thinking over to machines in the hope that this would set them free. But that only permitted other men with machines to enslave them”
~ Frank Herbert, Dune
Earlier in 2023 there was a bit more consternation than normal about a new Drake song. The track “Heart on my Sleeve” racked up millions of streams and views. The only problem is that it wasn’t actually done by Drake. The song was created using AI vocals likely as part of a publicity stunt and legal battle. The story blew up on social media with a curiously small number of pundits willing to directly address the elephant in the room: When did AI get so good that it could create fake songs that would actually fool millions of people and what are the implications of this development?
To what degree will human beings embrace consumption of AI-generated content? How much does it affect your enjoyment of a song knowing that it was spit out by an algorithm and not a labor of love of human songwriters, producers, and musicians? Apply this to movies, TV shows, artwork, YouTube videos, TikTok shorts, animations, novels, blogposts, (can you be sure what you are reading right now wasn’t created by ChatGPT?) — basically any and all forms of media.
Early evidence suggests human comfort with AI-generated content varies by context and genre. For example the porn industry has already seen rising demand for AI-generated artwork, X-rated animations, and even photo-realistic interactive digital companions. For many reasons, people do not apparently need their internet smut to be the creative work of live human beings.
But what about other contexts? Here’s a good thought exercise: Think of your favorite song. You’ve loved it for years, it is the soundtrack of some of your life’s most important moments, perhaps you’ve even wept to it time and again. Now imagine you learn today that that very song was not written by an artist you like, but rather was the output of a software algorithm. A programmer wrote a few descriptive words as a prompt and clicked “submit” and in 2 seconds out came the mp3 file. How do you now feel about that song? The same thought exercise can be done thinking of your favorite book or film or essay or even videogame.
How does your experience of that work change? Some people might argue that it changes nothing. “The song is the song,” they might say, and all that matters is the media itself. Many others would feel differently. Knowing that the work of art that moved them so deeply was not created by another person would cause it to lose something.
This is because art is meant to be expressive. “Expressive” here implies a subject — a person expressing something. When we love a work of art implicitly we also love the sense of oneness we have with the creator. We feel that creator speaking to us, we feel a connection, like some other person out there captured something important we mutually understand and value. This element, abstract as it may seem, is essential for most people I suspect.
This doesn’t mean we can’t enjoy content created by AI. Any of us can vibe to some software generated techno beat on a long car drive, or procedurally generated dungeons in an online RPG, or even a ChatGPT poem about Donald Trump. The feasibility question of AI’s ability to generate decent quality content is pretty settled at this point. The real question is how much the proportion of AI generated content consumption grows and what follows if it passes a certain threshold. What happens when most of the media we consume is no longer created by humans?
Imagine a world where anyone can create any song or movie or game they want just by a few AI prompts. What incentive remains for a company to hire a designer, music producer, artist, writer, or even programmers eventually? If humans in large measure decide that they no longer care about media being the creative expressive work of other humans, then what may follow is a world where human beings no longer make art.
Do we want to live in that world? How would art evolve under such conditions? AI-generated rock songs generated by feeding in a combination of other AI-generated rock songs ad infinitum. The clever reader might argue that the history of music made by humans is not essentially different, as styles and genres evolved over centuries of artists being influenced by their contemporaries and riffing on their ideas, from Beethoven to Eminem.
But the difference is that Beethoven and Eminem are both humans who had something to express about the world in which they lived. The new genres that evolved, from romantic classical music of the 19th century to dubstep in the 2000s, reflected shifts in human culture. What will a future of AI-generated music express and reflect? Perhaps it will aptly represent a new AI-generated man. In a smartphone-addicted age where so much behavior is dictated by social media algorithms, YouTube recommendations, the endless Twitter/TikTok content scroll, remakes and remasters across Hollywood and gaming — perhaps AI-generated art befits such a world more than we assume.
Whatever we believe the dilemma speaks to the importance of intentionality with respect to AI. We cannot simply default and hope the tech conglomerates drag humanity to an optimal future. We must take a stand and operate with some sort of principled framework. The writers’ strike happening now is a necessary example of this. Restricting the use of AI for creative work is a large component of the union’s demands. This is an essential point of grievance because it is not simply about an industry using technology to be more efficient but rather an industry fundamentally breaking its social contract with its customers by redefining the definition of “art.” Would media companies see the same number of streams, downloads, and ticket sales if viewers knew the movie they have been waiting to see was created by an algorithm? Likely not.
Technology ought to be a means not an end; it should serve us, not the other way around. Art is a fundamentally human end. If we take the humanity out of creative work we risk hurtling toward an inhuman future. For the singularity-loving tech-optimists, that is entirely the point. Taking humans out of thinking work is a feature not a bug. The alternative to that vision need not be regressive Luddism. Rather what is needed is a third way that creates space for innovation that serves humankind. We need leadership and vision to define that path.
