A whole new thing
Recently the Wordle charts in my Twitter feed have been displaced by grids of DALL-E images from this site, which gives predictive algorithms a few cracks at generating an image to match a phrase you submit. Typically these turn out to feature humans that look like half-rotten potatoes, or they look like abstract approximations of b2b ads from the 2000s. But still, it's hard to resist the temptation to find out how "the internet" sees some concept — how all the scraped and processed data from millions and millions of people's text and images averages out into blurry, dead-eyed blobs.
As novelist Robin Sloan pointed out, "nearly all presentations of the art produced with these models include the text prompt. The pleasure, it seems, is not in the image; rather, it’s in the spectacle of the computer’s interpretation." Or you might argue that it is in the funny things that people think to put in the text box. One can imagine all those human inputs being scraped to train another computer model that tries to generate clever prompts for the image generator to execute: Would that process manage to surprise and delight us more, or would the two AI products cancel each other out, like a higher-order adversarial network?
Generative models, Sloan's comment suggests, are just a medium, no more compelling in themselves than a blank tape. But they provide the occasion for recording different sorts of human interventions and expressive gestures. Anything presented as "AI art" is interesting only insofar as you can see through to who made the information that fed it, or how someone coded it, or how the inputs were massaged to get a particular outcome. Despite his previous optimism and excitement, Sloan concludes: "I don’t believe AI tools are useful for serious writers." That also seems to suggest that refusing AI tools will be a way for authors in the future to announce their seriousness.
One of the concerns frequently evoked (including by me) about generative art is that, since it is so dependent on old data, it is constrained to reproduce the past as the horizon of the future. AI models point to a fully administered world in which creativity is winnowed to choosing among prefabricated expressive options and nothing is left but the rearrangement of already existing concepts according to the lineaments of established genre models and the requirements of those in power. (In other words, the consumerist world we already have.) Everything "new" will exist only to testify to the fundamental unchangeability of things as they are, a social order that can reproduce itself endlessly in superficially novel ways, proving its capacity to absorb and neutralize any kind of change, resistance, or difference. Generative art models demonstrate that the status quo can be repeatedly repackaged as novelty for an eager base of perpetually distracted consumers conditioned to desire only things that come partly pre-consumed. (c.c. Adorno).
An earlier Real Life piece connected predictive algorithms with "vibes" along these lines:
The vibes are off, but they’re off fundamentally because they focus only on feelings and emotional connections that have already existed. They don’t provide or imagine pathways to new futures; they allow only for an understanding of what feels good or bad based on experiences that have already happened, things that have already been seen.
Michael Sacasas makes a similar case in a post called "We Are Not Living in a Simulation, We Are Living In the Past." He links the transmutations involved in such processes as AI-generated art to the possibility of being suffocated by past data:
We no longer encounter the past principally as a coherent narrative informing our present and future action into the world. The past, is now encoded in ponderous databases, and it can be readily and endlessly re-interpreted, reshuffled, recombined, and rearranged. This activity is what now consumes our time and energy.
That sounds a bit like some critics' complaints about postmodernism in the 1980s and 1990s, that culture had become detached from master narratives and was just a stew of arbitrary signifiers to be deployed by people who have mistaken solipsism and anomie for individualism and agency.
When you consider the TikTok that kicked off a lot of the "vibes" discourse — a guy drinking juice and skateboarding, set to Fleetwood Mac's "Dreams" — it's striking that there is nothing particularly nostalgic about it. The "vibe" is not a reactionary fantasy about the good old days, no hint that we are supposed to think about how music used to be better or more authentic or anything like that. Instead, the song is decontextualized in a way that is extreme even by postmodern standards: It is not used to signify anything about its history or legacy. You are not supposed to care about Stevie Nicks, or 1970s rock, or where "Dreams" fits into the canon or established genealogies of pop tastes, or what it says about someone as a music fan; it's just a sonic texture that the app made available. Its context is other TikToks and other TikTok users right at that moment. Given that its past is not particularly relevant to its present, it may as well be a new song.
That sets TikTok as a music-streaming platform apart from Spotify, whose architecture and interfaces still tend to situate listeners within the organizational structures the music industry has developed. Some playlists may be mood-oriented, but listeners can still navigate their way through the terrain of recorded music with some of the maps drawn up in the past and assiduously maintained by people who are invested in their accuracy. Whereas on TikTok, users aren't expected to navigate their way through anything — its algorithms take you by the hand and lead you, and you don't have to orient yourself at all to experience its entertainment value.
Since the context of what you are seeing is always "you," everything is in a sense automatically new merely because it is happening to you now. We aren't trapped in the past so much as trapped in novelty, deprived of the social wherewithal to develop concepts and build our own frameworks to organize and navigate what has been. TikTok's alertness to our desires doesn't make for narratives of cultural "progress"; it administers culture with a view that extends only as far as the next video it shows each individual user, and the trends that emerge speak to nothing more than that pacification. Maybe the claim that "People are getting tired of the 'TikTok music formula'" is better understood as "people are tired of there being no formula but their own dehistoricized manipulatability."
All of this makes me wonder, then, what to make of the fact that the nostalgia show Stranger Things can push Kate Bush's 1985 song "Running Up That Hill" to the U.S. top 10 in 2022. You could point to this as evidence that culture is being recycled, another example of retro revivalism. After all, it's not new or unusual for a song to be used in an ad or a movie and then return to the charts. You could declare that "old music is killing new music."
Or you could see this as the reduction of Kate Bush to a vibe and the negation of her legacy. You could argue rather that it marks the belated triumph of an irresistibly great work of art that can transcend whatever mediocre showcase it initially was given to become a phenomenon in its own right. You could say that it proves people are more "curious and open-minded" about music than the industry gives them credit for. Or that people are simply more malleable. Maybe it suggests nothing more than Netflix winning a skirmish in the endless war among media companies for cultural market share.
I first heard "Running Up That Hill" on MTV in 1985, and I couldn't have told you whether it was the video, the song, the cachet of MTV itself then, or my desperate need to signify to myself my difference from other kids in my high school that eventually led me to buy the Hounds of Love cassette. I still don't know if I even like it. I remember listening to it on a Walkman, fantasizing that somehow other people could hear what I was hearing and their being impressed. How could I adjust my walk so that maybe they would know? What would make them jealous of what they couldn't hear? This was social media to me.