Drommet red
In a essay for Outland about AI image generators, Kevin Buist argues that they change our relationship to the idea of genre.
It used to be that there was a limited set of words, phrases, and narrative archetypes that elicited the visual elements of a genre. “Cowboy” and “pirate,” for example, are words that conjure images in our minds that are defined by our shared understanding of their respective genres. But the list of words and phrases that can now be visualized in a general way is no longer limited to our loose list of genres. Now that we can elicit images from any expression of language, all possible phrases have become microgenres.
If genres once depended on mass media and took time to cohere and coalesce, AI image generators suggest that genres can now be instantly consecrated through processing power bringing to bear the concentrated attention of millions (in datafied, algorithmic form) onto a set of words to produce an endless supply of characteristic images related to them. Voilà: instant reification of an idea. An identity is born.
Genres don't have any particular relationship to mythemes or Jungian archetypes or particular socio-historical conditions demanding collective expression; they can now be understood as a spontaneous association of a phrase, any phrase, with a set of typifying images known to correlate with those words. Search engines — and the AI image generator could be understood as a form of search engine, as Buist notes — produce content "genres" on demand.
TikTok is another place where "everything is a genre now." Its algorithm is similarly credited with constituting ad hoc genres on the fly, not necessarily through users' specific requests but through its spontaneous production of mass audiences for emergent content themes. On TikTok, we are not operating the search engine, looking for what we want to know, but are instead being operated on by TikTok, and its algorithm's search for information relevant to it.
In the past, I've tended to think mainly about how TikTok's algorithm constitutes the individual user's identity, interpellating them as someone who has certain tastes and proclivities that it has magically uncovered. But the power of that illusion is interconnected with its powers of genre creation, its ability to instantiate trends. TikTok "knows who you really are" in the same way that it "shows what's really happening in the world." These "knowledges" have nothing to do with facticity but with the authority the app earns by holding people's attention.
This Errant Signal video (which I watched on this recommendation from Cameron Kunzelman) offers a lucid explanation of how TikTok produces genre, at the 10-minute mark and then again around minute 40. (It's a long video.) First, it details the vernacular sense in which TikTok has "sides," i.e. places where users can get stuck seeing the same kinds of content. These "sides" don't just reflect established niches or interests; they are brought into being around topics that the algorithm identifies as trending and begins promoting, which in turn sparks creators to make more TikToks appropriating or copying the theme, giving it more momentum. As the video explains:
The algorithm that drives the sides of TikTok doesn't use pre-programmed categories but dynamically invents its own as they occur. That way a breaking news story or a holiday or whatever can have its side emerge, bloom across For You pages for however long it's relevant, and then, as interest wanes, recede into the background. So when a person or account goes viral on TikTok, it's not just that lots of people are making videos about them or talking about them; people making all those videos makes the algorithm take notice and eventually gives them their own side of TikTok.
That is, TikTok uses individuated trial-and-error experiments on a massive scale to identify bundles of content that incite certain sorts of people to react and then pushes that content on them and everyone connected to them in some way. These bundles consequently become recognizable as instant genres, with a specific audience attached to them. That new concatenation of users can subsequently be redeployed to propagate further genres, as different content bundles are detected.
Given that both AI image generators and TikTok create ad hoc genres, one can think of them as analogous, with TikTok users functioning as the weighted parameters in the machine's code. TikTok could be seen the way Vilém Flusser saw photography, as described in Richard Woodall's account of it here: as a system that "recruits human beings to serve as its fleshy prostheses, employed to point the camera, press the shutter, and create photographs in such volume that a certain percentage will inevitably, by accident or by design, contain 'new information' to be absorbed into the program." TikTok users in effect work for its algorithms: The algorithm isn't "for you"; it's a technique optimized to compel you to act on its behalf. It works in concert with the site's other deinhibiting tactics, some of which are itemized in the Errant Signal video starting at about minute 25, under the guise of how TikTok makes it easier for users to "express themselves."
If "freedom of expression" is one way we understand the experience of this coercion, "genre" is another. A genre is not a set of tropes or themes but a concretized amount of attention that can be given a unique identifier — a search phrase or a -core or some other minimally viable unit of signification. When we recognize a genre, it marks our participation in a social experience, one we appear to be opting into. But it also signals how we have been corralled into a particular "audience commodity," to be exchanged for other people's profit.
In Aesthetic Theory, Adorno argues at one point that "the telos of artworks is a language whose words cannot be located on the spectrum; a language whose words are not imprisoned by a pre-stabilized universality." One might say that AI image generators preclude the possibility of such language — anything you type in yields a genre. But less pessimistically, one could see the AI as exhausting a particular idea's art-alien possibilities, showing us all the images that necessarily fail to express the idea aesthetically. It defines the field of possible art by showing us lots of non-art.
TikTok could be seen the same way: As its emergent genres calcify, we are able to concretely see something that is otherwise impossible to visualize: how the collective spirit that had been animating a particular phenomenon manages to escape at the very last second, just as the idea is being made into a cliché.