aesthetic accelerationism
a defense of using Claude to win a literary prize
The Many Deaths of Ocean Vuong
The big literary news of the month broke last week when a Claude-vestigation revealed that the Commonwealth Prize, something we were all definitely already aware of, was awarded to a writer who had clearly used Chat-GPT. The winners of the prize get published in Granta, an important literary magazine most famous for never accepting any of my work. Once the word was out, it quickly snowballed into a Dolezalean mass-schadenfreude event. Even a superficial look into prize-winning author Jamil Nazir revealed an AI-generated author photo and scads of LinkedIn posts, written by an LLM, in which he advocated for the use of LLMs. This shit, like ChatGPT, basically writes itself.
There is a familiar shape to how these scandals unfold. The first step is a close reading of the story’s most ridiculous sentences, the bleeding evidence of bloodlessness in the prose:
Naturally, in these posts, it’s important to establish first and foremost that a) I would never fall for this and b) my own writing is much better than this. Not that I disagree with the sentiment.
The second stage of the ‘discourse,’ as it were, is the rather obvious acknowledgment that it’s not unusual for this type of writing to win literary prizes:
Suddenly everyone has read Granta before
In the spirit of research, I read the story, even though I found it difficult to fight through one tortured metaphor after another (truly, almost every sentence ends with some histrionic or sentimental figurative language). It’s titled “The Serpent in the Grove,” and it concerns a vaguely Global South-y cast of characters in a vaguely Global South-y village and an instance of domestic violence that lingers in the form of a vaguely magical-realism-y humming well.
Interestingly, while the AI-of-it-all raised the flag, it was these specific story elements and above all the style that seemed to catch the most flak.
If you are not on the Literary Internet, and statistically you are not (can’t recommend!), you may not be aware that All is Not Well in the State of Denmark. It seems there is a new viral essay every other month or so arguing that, in the last two decades, the books industry was captured by a bodies-in-spaces form of identity politics that, among other things, prevented white men from getting published, explicitly sacrificed merit for the sake of representation, and just wasn’t very good. These ideas still get a good deal of pushback, but they are sayable in a way they certainly weren’t even five years ago.
These sharks circled around Ocean Vuong a few months ago, whose recent novel was gleefully panned for being, well, bad. Vuong has, in recent years, become an avatar for post-colonial aesthetic excess. Most famously, Som-Mai Nguyen’s essay “Blunt-Force Ethnic Credibility”flipped the script: rather than offering some radical critique of a white-dominated literary world, Vuong has succeeded by selling dubious interpretations of his culture to a credible white audience. In this light, his breathy reflections on language and poetry begin to ring false.
But Ocean Vuong is a too-big-for-his-britches (literally, in the case of one infamous photo shoot) literary superstar who hung out with Oprah. Jamil Nazir is a nobody (maybe literally — does this author even exist?) Nazir is not being attacked personally as much as his writing, the preferred style of the LLM, which seems to have been heavily informed by Vuong’s work and others similar.
Just a few years ago, such brazen attacks on the “Lyrical Post-Colonial Story” would NOT have flown, but packaged in an anti-AI broadside, people hardly bat an eye.
It is often difficult to criticize a genre, for you inevitably end up defending your own particular tastes, and every idiot knows that taste is just a subjective thing (taste is, of course, objective, but more on that later). It’s better to criticize, say, an MFA. To write an “MFA short story” is to write a worked-over, florid, limp narrative that’s been thoroughly focused-tested. But is it wrong to write polished and florid stories? Doesn’t matter — everyone hates the fact that you went to an MFA to get better at your creative pursuit.
And now, the critique of post-colonial literature has found its perfect tag: it reads like LLM prose. Just look, an LLM pastiche won a significant literary prize!
Sure, this style of writing will hobble on after the scandal, but I might hazard a guess that these blows are landing. My early impression is that “turning benches into men” has a memetic resonance that will damage the lyrical mode and how we read it. It is an aesthetic ‘quilting point,’ the literary ‘ick,’ wherein a previously floating subjective meaning becomes fixed. Who says that literary magazines aren’t relevant any more?
It’s worth considering this scandal from Granta’s perspective. One must imagine that they don’t hate the attention. I wonder how many more views this story received from their average fare? I imagine orders of magnitude. And yet the embarrassment of it all feels, at a gut level, much more serious. This contrast — the audience purchased at the cost of credibility — lays bare an essential dynamic in literary fiction: it never was about the audience size. Or at least, readership can be floated by credibility. People always cared (if they cared) about Granta insofar as it was prestigious.
But in light of this scandal, can Granta stay prestigious? What happens when lit mags in general lose prestige? Without an audience to fall back on, I think we can guess.
On Eels and Their Containment
Prestige is a more or less objective measurement, captured in weighted lists that consider things like distribution and major prizes (i.e. Pushcart, Best American Short Stories, O. Henry, etc.) won. It’s also an aesthetic category, though. There is a reason that literary stories sound like that. Conceptual metaphors and a heavily affected cadence just go with the territory.
An easy way to criticize “The Serpent in the Grove” is to gesture at a metaphor in the story and ask “what does that even mean lol.” What was Nazir getting at when he wrote:
“my father’s shadow slipping away down the corridors of the house as if it were a weasel, a ferret, or to employ a more appropriate simile, an eel in an inadequate container.”
Why is an eel in an inadequate container like a shadow? What even is an inadequate container for an eel? Moreover, how did the eel get there? Moreover moreover, everyone knows that lists of three, especially a self-contradictory list of three, is a dead-on-arrival sign of LLM-writing.
Of course, Nazir didn’t write that bit with the eel. That comes from Roberto Bolaño’s masterful By Night in Chile.
I’m sorry to do the old rug-pull bit but it’s the easiest (laziest) way to make my point, that responding to AI stories by saying it’s bad to do X or Y thing in your writing is failing the thought exercise. If “doesn’t make sense” is the line of criticism we are going with, what of Faulkner and Joyce?
I assume that many people will take the lesson of Grantagate to be to trim the fat of their fiction, to ditch the metaphor and the voice and tell it straight, Jack. Indeed, it seems these days that culture-makers will lunge for anything that might be the beginnings of a trend. In our age of nothing-ever-happening, more and more people are cultural prognosticators of the microscopic. What is the cool literary device of the summer? What color lenses on my character’s sunglasses? Are quotation marks passé? Is ditching them cheugy?
Aesthetics, of course, isn’t just whether poetry should rhyme or the color scheme of your living room. As a field of study, it is more concerned with the interpretive or affective lens through which humans make art or find art meaningful. Art movements arise in particular contexts, and sometimes when the contexts change, the movements lose their relevance and bite. Vuong did matter in a particular moment (think: Obama presidency, Buzzfeed, Colbert was still funny). Now, I’m sorry to say he’s fucking chopped. There but for the grace of god go I.
I love the book Jack the Modernist for how it lays bare this principle, the relation between aesthetics and society. Robert Glück writes:
Lack of meaning is exhausting. I never realized that these neighborhoods and childhoods were the eye of the tornado that the U.S.A. is to the 20th century. L.A. was so undifferentiated that movement seemed inadequate, remote: setting out on a walk like they do in books, returning diminished and helpless, miles of tract housing defeating us at every turn: at the vanishing point a couple of treeless kids in bathing suits would double and shift like flames. Later, a hundred thousand electric garage doors opened to receive the unwelcome fathers. Brian and I assuaged this memory of suffering by embracing it—the embrace conferring a human scale the original lacked. Then we offered ourselves—in the spirit of revenge—as myths produced by a dead society.
You end up with the art the age demands. For Glück, his post-modernist embrace of campy pop culture is a tactic to create meaning where none exists. This was of particular importance to gay men in the early 80s, who did not yet have a home in the culture. His aesthetics is more than just determining the correct number of adjectives in each sentence. Although he does write incredible sentences.
The Granta scandal seems to me like the first major aesthetic clash in the AI age. Some of the old interpretive models, having withstood a period of erosion, are toppling. Here, AI emerges not as a creative force, but a destructive one. Claude chews up language and grammar and vomits its house-style back onto you. In such a system, there is a demand for an aesthetics that resists the machinic enzymes, a cold diamond of prose that proves indigestible to LLMs.
LLMs, perhaps unwittingly, function as effective satire, seeking out aesthetic modes that no longer serve us and reproducing them in forms we cannot take seriously. I wrote previously about anti-aura, the rizz vaccuum, which seeks out art to make it meaningless. It is the job of the artist to respond to such a technology, to neither ignore it nor accede to it. One other object lesson of the Granta fiasco was the universal vitriol and mockery this writing received. If nothing else, that should inspire some hope that people won’t go quietly in to the LLM night.
In the final analysis, it seems that Claude has done his job here. More than a thousand think-pieces ever could, he discredited an entire style of writing. One is tempted to take the accelerationist posture, using AI like a corpse-sniffing dog to identify the styles we are better to dispense with.
Scattered thoughts on artists who Do This:
It is maybe too pedantic to list off the qualities of writing that lend themselves to unintentional AI parody, but the “lyrical mode” of sentimental metaphor and neat parallelism seems like a fine starting point. But it’s also insufficient. The problem is more structural than the sentence level.
One of the simplest paths forward might be to return to the core function of stories. Not ‘short stories’ as a contemporary genre, but the timeless art of narrative. Two Substack writers I read who I think have their finger on this pulse are Ben Sims and Naomi Kanakia (I particularly enjoyed this essay and story by Naomi). Ben, I might add, writes with some real Cockney flare, a reminder not to throw the baby (literature in a vernacular style) out with the bathwater. By taking a Brechtian or dialectical approach, one might throw cold water on the reader and escape the trap of dissecting the figurative language for meaning. Narratives that rely on the dual pistons of trauma and epiphany have grown stale. I might recommend my Baffler story, if I may be so bold, as an example of fiction that hews to narrative qua narrative. I strongly believe that Claude could not, would not, have written anything like “The List.”
The other mode that has struck me recently as being particularly timely are the more intricate or narratively nested structures. How refreshing it is, in the era of Claude, to read Bolaño, a writer deeply attuned to structure. One must be willing to venture deep into the labyrinth to open the treasure chest, which most likely is empty anyway. Stories are told within stories, and whether the reader chooses to take those tales at face value is a fraught choice. The structure informs and is informed by the political context of his work, the agonizing reckoning with evil in the world and in ourselves. In Bolaño’s aesthetics, surreal images don’t read as vapid flourishes, but interpretive keystones for his narrative cathedrals.
There are two other writers in particular who have moved me recently with fiction that seems resistant to LLMs (both on Connor’s recommendation. Thanks, Connor!). Yasunari Kawabata’s celebrated Palm-of-the-Hand Stories have that same nested quality, stories relayed through complicated chains of relation and observation. Yet Kawabata fixes these intricate tales in the short frame of a page or two, like setting a gemstone in a gold band. The other is Jesse Ball, whose novel The Way Through Doors takes a more Escherian route. There is the same self-referential quality, in which the story you are being told is constantly interrupted by another, even more dubious tale, until at last a door or window opens and you find yourself back in the proper story, three levels up, for a few pages, only to promptly descend again. I should add that both of these writers know their way around an image. The essential thing is the relationship between these intricate structures and the images that serve them and are served by them. I doubt an LLM would be able to build the schematics.
(Am I just telling you the books I like, which are better than the books that other people like, and ascribing some messianic aesthetic manna to them where none is warranted? No. No, I’m not. As I said before, taste is in fact objective, and if you don’t like Bolaño you are, in fact, wrong.)
Maybe Claude will learn to do these modes too. Meta-narrative is, ultimately, a sort of complexity value. I kind of doubt it, but maybe ‘more compute’ can solve it. In either case, it is precisely the field of aesthetics that give humans an upper hand. Our written styles can evolve to suit the world we exist in. In some instances, we co-create that world. LLMs are doomed to only react; as they need the raw material to create the pastiche. True creativity, true originality, remain human domains.
Reading back through this, I feel compelled to state the same idea more strongly, lest this seem like so much cope. It is not enough that the human capacity for creativity is a permanent upper hand against Large Language Models. It is the duty of writers to articulate an aesthetics that can meet the moment. Vuongist stories fail because of the vapidity of the prose and, in fact, their broader ideological project. The fact that an LLM can succeed at recreating it is just a litmus test. The writing is bad whether or not it comes out of a computer.
It’s not to say it won’t be a difficult moment to survive. For lit mags to lose their prestige would be a blow to anyone who cares about narrative or fiction in general. It’s a time for strong criticism — of our selves and each other — and robust community building. We need new schools, new tendencies, and we need them yesterday.
In the meantime, I will be submitting a story to Granta when they open to submissions in June. Because I’m a generous guy, and I hate to see a friend suffering some bad press. I might have higher standards than all that, but if I can help them right the ship with some fiction that does something, I’m willing to make an exception.