Last year, Workday, a tech firm whose artificial intelligence tools are used by managers at over eleven thousand organizations worldwide to sift through hundreds of online job applications, was sued for discriminating against applicants. A Morehouse graduate, Derek Mobley, with nearly a decade of work history, was joined by a handful of other plaintiffs arguing that Workday’s algorithmic hiring tools discriminated against applicants based on “immutable characteristics” such as race or disability. Mobley and the other plaintiffs’ claims will be woefully familiar to anyone who has been searching miserably for work over the last two years. Applications sent in at almost one in the morning, only to be rejected within the hour, unseen by a human being. A person of color can apply to hundreds of jobs at Fortune 500 companies, yet because A.I. tools, like a transmogrified form of autocorrect, replicate patterns without regard for history or ethics, a bot may eliminate cover letters that do not fit the existing dataset whenever a socially dominant ethno-racial group, a gender, or sexuality is overrepresented in applications. A predominantly white workplace might, inadvertently or not, use a bot that immediately rejects applications with stereotypically non-white names like “José Sanchez” or “Trayvon Washington” or “Muhammad al-Khalidi.” These bots calcify existing hierarchies, and these technologies enshittify an increasing number of social practices from job-hunting to online dating, and even war-making.
Are algorithms racist? Is ChatGPT schizogenic? Is A.I. art actually art? Is an influencer a worker? Whatever happened to Jezebel and Tumblr, and just what the hell were those things, anyway? These questions are dealt with in Amateurs! How We Built Internet Culture, and Why It Matters. Many theory-inclined Marxists—especially those in their twenties and thirties who can scarcely remember life without the whole Internet at their fingertips—would be intrigued by a theorization of the stuff that has unthinkingly reared them. They’ve had platforms such as Instagram since childhood. Platforms such as Twitter, once-cherished, are now enshittified by corporations. Most pressingly, large language models (L.L.M.s) are developing before our eyes.
In the hands of Walsh, the “amateur” of the digital age is many things: an artisan made low by capital’s inauguration of the so-called “Web 2.0” era in the early aughts, existing as something between proletarian and petit-bourgeois. It is this amateur aesthetic that Walsh is interested in, the sort of stuff that professional artists and critics snub. Walsh gravitates to a potentially kitschy aesthetic in which “artifice that imitates artifice, that produces apparently nontransgressive work that does not feel traditionally risky.” She is interested in a “low-class art” that “rehearses non risky emotions, or rehearses them in highly codified forms that seem, on the surface, to neutralise any risk involved.” When she claims that the left has been “shaken” by the growth of Internet aesthetics, ceding that ground to the right, what is she seriously proposing instead? If the White House wants to post AI slop depicting kidnapped Latinos, I don’t think we need to get in the habit of posting “AI slop but make it leftist” in response. Go to school. Pick up a paintbrush. Hang out with other artists. If it’s not broke, why fix it?
Amateurs, Walsh argues, can produce art of sorts. Or, at the least, users have an aesthetic. She develops this claim across a chapter-length analysis of amateur-generated Internet aesthetics, taking as her point of departure a humble and ubiquitous meme: the LOLcats inanities you may remember from the late aughts. From this, Walsh offers a theorization of what makes Internet aesthetics a break from earlier conventions: the collapse of the divide between producer and consumer. Meme-maker and meme-enjoyer are conjoined into a meme-factory that has “no beginning or end and its rules transpire through practice,” proliferating endlessly, liberated from the constraints of the enfleshed. In a provocation sure to spark eye-rolls, Walsh considers the potential, yes, radicalism of making memes on the Internet, for the way they upend the relations and norms of the art world through their “physically dispersed networks” governed by anonymity and ephemera.
Free-wheeling and virtual, the Internet allows the consuming critic and producing artist to be one and the same cycle of clickbait. It is everywhere, everyone, and all at once. Valuation in this online aesthetic universe, as seen through memes, is propelled not by originality, but by “proximity” to the “source or context of original production.” A valuable meme is one that is popular and flexible enough to proliferate. This unpredictable and unstable relationship between authorship and circulation is what makes memes and other internet aesthetics so uncannily unsettling, yet so fun and potentially radical in their chaos. Zoomer and millennial humor is often said to rest on the absurd zaniness of memes, which, as Walsh argues (following Sianne Ngai’s theorization of contemporary aesthetic categories), lies in a “performance of precarity in the face of capitalism, which is something to do with production.” In other words, imagine me and the gays—grown thirtysomething men, mind you—at the bar quoting and riffing off the same thirty-five to forty-seven Real Housewives or other gay brainrot memes to each other to the point that they have become verbal tics … Go piss, girl!
Digital aesthetics are not just symptoms of precarity—they practically melt one’s face off. Whereas the avant-gardists of the twentieth-century stood athwart Fordism and its mushrooming sameness, seeking, through “recuperation” of the mass-produced—think Warhol’s Campbell’s cans—to make art, “memes shift the role of artist from individual curator of the mass-produced to mass producer.” Walsh argues that it is this very ungovernability that made the NFT craze come and go so quickly. Enabled by blockchains that could trace authorship and ownership, these tokens, often made by professional artists, tried to resolve the contradiction between infinite oversupply and real-life financial value. Instead, they oversaturated the market with a dizzying volume of tokens of unimaginably inflated prices and thus collapsed the market under its own weight.
Luddites will be heartened by the sordid evidence marshaled in the fifth chapter, “The Sun Dall-E Painters of the Internet” on L.L.M.s and other advanced A.I. tools such as DALL-E itself, which, much like text generators, is itself prompted by texts. Walsh argues, quoting the scholar Hito Steyerl, that the algorithms behind the discrimination alleged in the Workday lawsuit are not an unknowable black box. Rather, they’re a “white box”: a social filter that mirrors preexisting hierarchies and structures. As Walsh puts it, “It’s not the machine that’s the problem; it’s the people running it.” When fallible creatures of history such as ourselves are the ones writing the code, collating the datasets, and building the algorithms, whatever is underrepresented in the makers’ own data or knowledge shows up missing in the bots. Walsh points to a case of a “primitive” prediction tool that precedes ChatGPT—your run-of-the-mill autocorrect—that, in an article about Ethiopian cuisine, changed the words “doro wot” to “door wot” and “awaze tibs” to “aware ties.” It was an error scandalous to me as both an injera fanatic and a honey wine enthusiast. What Silicon Valley lefties such as Timnit Gebru have sounded alarms over is the now emergent truth that AI functions as “propaganda for the ideology of the predictive,” reflecting existing inequalities back at us. Garbage in, garbage out.
Understandably, the very idea that AI art is “art” draws snarky denunciations online and offline. And for sound reasons. Walsh, almost wanting to pop the bubbles of stuffy lefties who denounce it all as amateurish crap, brings us right to the edge of saying that slop is art, but stops short of a decisive judgment. Nevertheless, Walsh, in what feels like an inadvertent and roundabout fashion, lays out an analysis of internet aesthetics that gives us a good idea of what it is exactly about AI art that seems … not like art. Walsh is sympathetic to the idea that AI can produce “art.” or that amateurish things like memes are an art form in their own right. Open-minded people can surely warm to a definition of art broad enough that everything from cat memes to the slop that has flooded every social media platform (the “Shrimp Jesus” phenomenon of 2024, say) can be called an art form. But an overly open mind starts to lose its grip on discernment. There might still be such a thing as “good” art versus “bad” art. For one, even though Picasso stood on the shoulders of giants, AI is nakedly derivative in that it points not to innovation, but toward endless synthesis. AI art flies in the face of the idea of the lone artist, struck by the divine and grasping for the sublime in solitude. AI art is “openly a group effort” that coagulates into a “mashup of existing art.” This is a part of the uncanny look of AI art, eerily looking “not like art but like a commodity … too deviously and carefully aesthetic–too referential to other modes, too slick, too flattering–to be art.” Art is “notoriously hard to define” as Ted Chiang has argued in The New Yorker, yet it can be generalized as “results from making a lot of choices.” A novelist decides words sentence by sentence, a painter their strokes and colors centimeter by inch on a canvas, and a composer note by note, second by minute. Even designers of animated films or video games are making decisions down to the pixel. You ask a bot to do something as simple as write a paragraph about a summer’s day or generate pictures of the items on a menu just to get, in the former case, prose so reductively boilerplate that it’s risible. As for the latter, I do not know if readers have seen this, yet there is a rash of bodegas and other eateries in Brooklyn hoisting up AI-made chopped cheeses and tacos that look like eldritch nightmares certain to give one trypophobia. Cthulhu-kibble. Does AI produce interesting things? Interesting as far as looking at a snuff clip is interesting, sure, but is it good art? That Walsh waffles and dances around that question is a disservice.
In the free-floating chaos of the digital epoch, particularly in its Web 2.0 phase, surface reigns supreme. A hall of mirrors entraps us, even at the level of taste and aesthetic. Walsh identifies the “core” aesthetics of our time—everything from cottagecore to academia core—as products of this. You don’t have to have actually attended an Ivy or Oxbridge to buy into the preppy academia core aesthetic, light or dark. Don’t know who Dostoevsky or de Beauvoir are? No matter! And you certainly do not have to be living on the prairie, or even have pockets deep enough to afford the “cottagecore” look in the way a Nara Smith can. You don’t even have to be a tradwife with neo-Nazi leanings! Pick and choose to your heart’s content. It’s all about you, you, you. This kind of thing is, per Walter Benjamin, less “folk art” and more what he, and Adorno after him, condemned as “the production of kitsch”: the “alienated, commodified” slop that capital produces, shorn of originality or ties to flesh-and-blood communities of trained and exhibited artists. Gone is the romantic idea of novelists who haunted Parisian salons and Manhattan cafés.
Walsh names her sixth chapter “2020: Wikidentities” to make sense of a paradox in the identity politics in that period: Even as everyone was online, social media also promised the possibility of “disappearing, of deleting your account, of having an alt, of not continuing, at least online, to be that thing your offline circumstances dictate.” In olden days, you identified as a communist or a fascist because “political identities were often judged according to scale, valued by the proportion of people” who adhered to them, thus killing off the growth of the most batshit ones. You went to meetings, read the Grundrisse with your comrades, got your co-workers in on the action, sent your children to commie summer camp, lived in the social housing project that your party’s elected officials got funding to construct, and went to the demonstrations on the weekends, bricks in hand. Say what you want about the Cathars or Sendero Luminoso, at least they commanded actual armies of believers! What do groypers and furries have these days?
Walsh has given us a text about as convoluted as the World Wide Web itself, humming with asides and tangents and flourishes that, while often jam-packed with compendia of morsel-like factoids and streaks of insight, could have used a bit more structure. Despite the academic language, the argument lacks scaffolding, and terms go undefined. Chapters are hard to summarize, resisting any boil-down. Cohesion slips. A thematic rather than chronological ordering could have aided readers. She does not offer examples from real-life—or “meatspace,” as she and others call it, that point to the politically efficacious, socially weightful, or culturally generative aspects of online culture and aesthetics. For example, in the “2020: Wikidentities” chapter, Walsh walks us through the explosion of countless elective identities and the heartfelt politics attached to them in the digital age. Think affinity groups for graysexual Furries or Solo Polyamourous Hijabi Amputee. Walsh sets up aesthetic movements like the Surrealists against web-based ones like “dark academia” mood boards, the former anchored in flesh-and-blood artists and the supposed “constraints” of “top-down dissemination” of books and films. She points to Online-Americans of color who subvert and play with the unexamined imagery of Oxbridge and Ivy academia as historic reserves of bourgeois white men. Cute, sure, yet I’d wager more people can recognize a Pollock or a Mary Shepard painting than have the faintest idea what separates light, gray, and dark academia is debated in the darkest corners of Pinterest and Reddit. If anything, a discussion of the exhaustion with online aesthetics or an evaluation of the effect of increasingly narrower and narrower identity-politicking on the post-Occupy left, would have been more interesting here. Walsh provides a fun layup, though.
Amateurs often feels like an attempt at valorizing the most ineffectual, most cringeworthy parts of our digital existences: a near-totally monetized slice of our lives controlled by platform capitalists in cahoots with the surveillance state. If anything, social media’s fragmentation of identity has driven the political fracturing of the uprisings and upheavals of the social media age—from Occupy onward—as Vincent Bevins has told it. Just as profiles come and go, so did the Occupy encampments, the hashtagged demonstrations of the Twenty-Tens that followed—all to the right’s benefit, which is fueled by the social atomization today’s digitized social landscape necessarily produces. The Internet era has birthed, following Anton Jäger, a hyperpolitical one. For the far-right that has consisted of sporadic lone wolf mass shooters fed on diets of Reddits and manosphere podcasts, and not the mass organizing of squadristi on city streets. On the far-left it has produced eternal squabbling in tweets and reels, pundits and anonymous trolls by the thousand typing in all-caps about whether or not Miley Cyrus twerking at the V.M.A.s was white supremacist, if Beyoncé earning another hundred million was the Combahee River Collective’s wildest dream, and other matters of consumption. Debates strategizing over how to build durable left institutions in workplaces or in the electoral realm toward collective goals of taking power were marginalized in a culture marred by the spellbindingly effective smartphone boom. God forbid you suggest on Twitter that you should do the patient, slow work of reading Capital or knocking on your normie neighbors’ doors to help get the union nurse that lives a few blocks away on the city council. Perhaps it had to take a steadfastly offline septuagenarian democratic socialist in 2016 to get us back on track, whipping millennials raised online into shape: to touch grass and do the real-world organizing that eventually led us to the election of Zohran Mamdani and other leftists.
This is not to discount the importance of the online world to, say, my own political and even sexual development. In high school, as I was coming out, chatrooms and Facebook discussion boards were where I cut my teeth, debating strangers about how stupid Ayn Rand was, or learning what frotting was, or who the hell James Baldwin was. It was not until I got to college, and then moved to Brooklyn, that I joined the D.S.A. and found a critical mass of gay friends that I really flowered. Amateurs often read like one long attempt at arguing for a non-judgmental and broad open-mindedness that flirts with abandoning one’s good judgment. It is a book that should be read as a compendium of astute descriptions of what digital culture consists of. Like, what exactly is an influencer and what is the labor that they do? What are the gender politics of the personal essay moment á la Jezebel and Jia Tolentino? And with rose-tinted glasses I am sure you can find all sorts of things in Internet culture to describe with fuzzy, feel-good leftist buzzwords like “communal” or what have you.
Yes, a meme from a Reddit page or an AI-generated image can be called the product of collective endeavour, just as the bomb that dropped on Hiroshima was, but who cares? It is hard to see how, on balance, anyone or anything benefits, much less the furtherance of civilization, from so-called “art” generated by platform capitalists and AI companies working hand-in-glove with ICE and Israel. Whatever Walsh says, there is no viable alternative the internet offers to art and politics. Though I suppose it’s entertaining to think otherwise.
José Sanchez
José Sanchez is a writer based in Brooklyn. His writing has appeared in n+1, The Nation, Jacobin, and elsewhere.