With thanks to Amadeus Code co-founder Taishi Fukuyama, for his thoughts on AI, art, and value. This originally appeared on Music Tectonics.
The digital revolution has been an explosion in the transmission and reproduction of ideas, words, images, videos, processes, and (coming soon) experiences. It is the latest moment in a longer arc, one that reaches back to the first technological ways people began copying works of art (and this includes music) swiftly and en masse.
Yet even when reproduction became industrial in scale, it was finite. Only so much could be manufactured, stored, shipped, and displayed for sale. When reproduction is nearly infinite online, when even the process of production can be reproduced ad infinitum via algorithms, what value does a work of art, an idea have?
First, a few assumptions that will help the discussion below make more sense. Art has value, and by that I mean both profound human benefit to its creator and savorer and more tangible exchange value. Because we’re talking business here, I’m going to focus on the later part of value (one we Americans often confuse with the former spiritual bit). And for the purpose of this discussion, I’m lumping music, theater, literature and its offshoots, and all kinds of visual media together when I talk about “art.”
Value in art has a history, one that has shifted notably in the last 200 years. In the misty days before about a century and a half or so ago, the era before mechanical reproduction of images and moments, a work of art was a discrete, unique object or performance. Its value flowed from its “aura,” as cultural critic and thinker Walter Benjamin dubbed it. Benjamin ties this halo of meaning and value to distance and, indirectly, to access. The distance might be spiritual or psychological–some aspect of religious experience or ritual that did not allow most people to approach a work of art–or socio-economic–only someone really wealthy or connected to the wealthy got to see, hear, read, or watch art.
Aura exists because of restriction, and restriction applied to objects and performances where the combined skill of the practitioners and the expense of the materials reflected the glory of the divine and/or the prestige of the work’s patron. In many cases, that patron might be the sole viewer of the commissioned work. If you were lucky, you got to view the nobleman’s personal collection or hear a private music performance. Otherwise, a house of worship might be one of the few places outside of public architecture where ordinary people could experience art.
This aura created value.
That aura got sidelined in the successive waves of machine-enabled reproduction, though it lives on in auction houses and exclusive venues. We could start our whirlwind tour of reproduction tech with movable-type presses and improved etching and engraving techniques, though they are extremely labor-intensive. Reproduction really took off several centuries later, as technologies started hitting fast and furious: first lithography and photography, then film, then sound recordings. These reproduction methods scaled well and relatively quickly. They consolidated into firmly framed media. Some technical aspects changed–the addition of color in film, long-play vinyl records–but the major changes came in transmission (cinemas versus in-home TV viewing, Victrolas vs. radios).
Transmission finally became its own method of reproduction, especially once it moved from ephemeral broadcast to documented phenomenon on the internet. Over time, this online reproduction became next to infinite as internet usage became fast and ubiquitous. There is next to no limit on how many times an image, sound file, streaming video, or other work can be reproduced. The “object” of this work is a mere package, a conveyance. Uniqueness feels absurd in this new digital world, the peculiar hobby of the elite investor.
We are moving now, thanks to algorithms and machine learning, into new reproduction territory, where a machine does not just copy the final product in some way, but copies the production process itself. AI can generate a song, a poem, a short story, a video, a news item. It can go haywire, but does a passable job. It is rapidly improving.
Mechanical reproduction made art unrestricted, something cultural critics in the early 20th century grappled with extensively. Some, like British anarchist Herbert Read, felt this was a good thing, that everyone should have full access to art and the role of artist, as art was useful to everyone. Others were more ambivalent, to a point that feels quaint in hindsight. Critics missed the aura’s ties to the object, and across the political spectrum, thinkers fretted about how notions like authenticity would fare if everyone has a cheap copy of a painting. Now that there were a hundred thousand images of that painting, the image itself became worth far less (though the physical paintings have retained enough aura power to stay expensive as all get out).
Now we’ve reached the point of infinite reproduction. Nothing has value, even the most beautiful sounds and images, cry some. All is data, broken down and rehashed according to rules, then passed around willy nilly. Machine learning promises to drag the asymptote of value ever closer to zero, drowning us in a wash of meaningless but precisely targeted works, a flood of mediocre experience without artistry, with no agenda beyond commerce. AI will bring the anti-aura apocalypse. We will lose art.
Yet digital culture is filled with highly valued works, whether we’re talking viral videos; the still images in popular social media accounts; the songs of sleeper-hit indie musicians with cult followings; or the centuries-old work of art that is a popular icon or recognized symbol, a painted meme. The value in our age flows not so much from restriction and aura, as from access and audience. Value is determined by reception and engagement, not production, a truism most tech entrepreneurs know well enough to roll their eyes. Yet Duchamp had the modern condition of art right: Art comes from context and audience reception, not the urinal on display. The object has no value, but its reception does. It is value.
What does this mean for art in our age of infinite reproduction, as machines move ever deeper into an automated creative process? We may throw off the yoke of aura and cease to need originality, uniqueness, and authenticity so badly, as we listen or watch or experience something. We may not need the myth of the genius to feel deep meaning and worth in a work. We may meet the creator as an equal.
There is the potential for nasty cheapening of aesthetic experiences, one that Aldous Huxley would recognize–as well as a powerful, accessible, mind-blowing opportunity to have mass creation, mass creative empowerment, in the place of mass reproduction. The tools are being laid out for human creation and collaboration on an unprecedented scale. These tools could alleviate the drudgery of production and unfetter the creative inner worlds of us all. Even if we can’t carry a tune, suck at drawing, or can’t spell for shit.
Where do we go from here? It depends on what we decide to value.