articles
issues
support
about
“AI” is a Candy House Jennifer Egan’s prescient reclamation of human sensemaking
Caddie Alford & Chris J. Thomas jul 2026 · review
Photo: Aagnverglaser; image cropped and colour-adjusted from original. CC BY 4.0.

Jennifer Egan’s 2010 A Visit from the Goon Squad is widely regarded as one of the greatest contemporary novels. It was awarded the 2011 Pulitzer Prize for Fiction and ranked 39th on The New York Times list of 100 best books of the 21st century. For many, the impact came back to how it pushed the technical boundaries of literary storytelling. Most memorably, one chapter is composed as a PowerPoint presentation. Readers were dazzled, not yet primed for a near, corporatised future that would standardise the slideshow into the slide deck.

Building on this success, Egan collaborated with The New Yorker to publish the first short story serialised on Twitter, tweeted out over nine nights in the requisite 140 characters. This was 2012 Twitter; the only people who would see these tweets (in reverse-chronological order!) would be followers of the @NyerFiction account. A couple years prior, Twitter had changed its composer prompt from “What are you doing?” to “What’s happening?” This, as media studies scholar José van Dijck points out, reflected the platform’s intended shift from “hosting […] interactive personal talk” to incentivising “news and information that also goes beyond the personal.”1 Under this new paradigm — The Tweets Must Flow — Twitter began to position itself more firmly as a “neutral” hub for uninterrupted speech.2 And still, the 140-character constraint endured, exacting mini-texts that often took on the shape of polished aphorisms. Popular personal tweets sounded like something you didn’t know you thought until you saw someone else’s pithy articulation. Take this famous Roger Ebert tweet from 2011: “Never marry someone who doesn’t love the movies you love. Sooner or later, that person will not love you.”3 Retweet if you agree.

Egan recalls that the idea of tweeting a story came before the actual story. Reading that story, “Black Box”, you get the sense that she relished learning how to speak Twitter: each segment exudes fresh appreciation for emergent generic forms. “Black Box” takes us into the mind of a female cyborg spy working as a special operative for the state sometime in the future. Most of this labour, we learn, revolves around manipulating rich and powerful men. Told in the second person and refracted through the medium of 2012 Twitter, each despatch is chilling in its detached intimacy:

Even a powerful man will
be briefly self-conscious
when he first disrobes to his
bathing suit.4

These aphorisms seamlessly blend in with the politics of a faceless government, part truism and part directive:

Remind yourself that you
Aren’t being paid when he
Leads you behind a boulder
And pulls you on his lap.

It is no surprise that the original title of the piece was “Lessons Learned.” Reflections abound in the spy’s mind, creating what Egan describes as a “didactic quality” not just to the story, but by extension to Egan’s interpretation of Twitter.5

From here on out, Egan dwelt with aphorisms, cultivating a fascination with the kind of wisdom that can only come from a polyphony of lived experiences. As the years progressed and the harms of social media platforms became all too clear, she shed most of the techno-optimism that you can still feel in the form of “Black Box” and embraced, instead, what the Digital Inquiry Speculation Collaboration Optimism Network (DISCO) collective might call technoskepticism: a posture toward digitality that “mediates between the two poles of optimism and refusal”, and “makes space for ambivalence, for the paradoxical cohabitation of joy and doubt, curiosity and caution.”6

The result is the novel The Candy House, which we argue offers a necessary counterweight to the hype-driven “AI” summer that we’ve all been suffering and cleaning up after since November 30, 2022.7 The Candy House — a loose extension of the world of Goon Squad — was published in 2022 and was perhaps a victim of its predecessor’s success. Big Tech didn’t have the same nostalgic charm as 80s punk rock. A lot has changed in four years, however. At the time of publication, the book offered a perspective that many of us were not yet prepared to (forcibly) receive. When it came out, much was made about the titular candy house and the lessons imparted from such folktales as Hansel and Gretel: nothing is free. Technology is a series of giving-and-taking, but mostly taking. And yet, we received this lesson like an abstract warning because opting out still felt possible. It is a completely different thing, however, to read this novel at the same time that Big Tech is, in Matt Seybold’s words, giving up on an “organic adoption” of their “AI” products to compel “a new dream of forced adoption”, mandated by “government and managerial coercion.”8 In 2022, we were still sharing memes about Zuckerberg’s Metaverse announcement; but in 2025, Anthropic signed a two-year contract with the Department of Defense/War; and in 2026, the U.S. military used Anthropic’s Claude in their seizure of Venezuelan President Nicolás Maduro — the definition of technofascism. We are no longer happening upon the candy house as much as living there now.

The Candy House orbits around a Zuckerberg/Altman-esque tech mogul — a figure Egan understands better than most, having dated Steve Jobs as an undergraduate — who creates a functional “metaverse”: Bix Buton’s Own Your Unconscious allows users to upload their memories, and in turn access other users’ memories. Each chapter focuses on a different character and their relationship to Own Your Unconscious. Much like the U.S. context across social media’s history of platformisation, very few in The Candy House technically opt in. Being a user is less of a choice and more of an imposition through the usual power dynamics, casual marketing rhetorics, gradual diffusion, and integration into such major governmental institutions as the justice system. Egan’s candy house anticipates what Big Tech tries to tempt us to believe about “AI”: with these technologies, with their products, we can be anything and know everything. Egan’s dizzying and experimental world-building strategically illustrates an intensification of capture, enclosure — and, eventually, of paranoia. As a character in The Candy House puts it: “knowing everything is too much like knowing nothing; without a story, it’s all just information.”

This is a novel of algorithms, just not in the way you might think: through both content and form, The Candy House reveals both similarities and crucial divergences between, on the one hand, the stuff of aphorisms and clichés; and on the other, the logics of statistical probability and surveillance. Egan ended up remediating “Black Box” as a chapter for The Candy House titled “Lulu the Spy, 2032”. It remains in its original tweeted form, but the association with tweets is no longer perceptible: thirteen years later, “tweet” has become what Egan’s extended Goon Squad universe would refer to as a “word casing”, or “words that [have] been shucked of their meanings and reduced to husks.”9 Think, for example, of this opener: “I hope this email finds you well.” That phrase is so common now because of generative “AI” that it’s become a word casing in sentence form.

In contrast to Goon Squad’s ephemeral word casings, The Candy House introduces “stockblocks”: those enduring clichés that are so cliché that we don’t even recognise them as such. Stockblocks are film clichés made mathematical. Chris Salazar — the protagonist of the aptly named chapter “i, the Protagonist” — leaves his editing job to work for SweetSpot Networks, where he is tasked with creating “algebraisations” of stock elements in film. For example, the classic drink-thrown-in-the-face-of-the-protagonist might be converted into:

a (+drink) x (action of throwing drink) = a (-drink) + i/2

“Having a drink thrown at you is humiliating,” Chris reasons, “which is more likely to make i feel reduced, or i halved.”

We all implicitly know stockblocks as the cues that tell us how to feel about the narrative moment. We’ve seen “Protagonist Hits Bottom Alone, at Night, on City Streets (with Soulful Music) [3Aixd]” and “Blurred Faces Lean Over Protagonist, Gradually Sharpening [3Axip]” countless times. Our reactions to scenes with stockblocks are somewhat pre-programmed; they resonate because of how we’ve learnt they resonate. Stockblocks are much like stock characters — think: the goofy sidekick — and like stock characters, they take care of some of the heavy lifting. The more stockblocks a movie or TV show incorporates, the less we have to interpret. It is for this reason that we sometimes hear the argument that “AI” can generate TV shows and movies: vary existing stockblocks enough, and you’ll probably elicit the reaction appropriate to the genre.

As we see it, it is The Candy House’s insistence on tropes, stereotypes, and human-coded algebraic formulations that turns it into a ballast to Big Tech’s intended colonisation. Such resources — the means to “predict”, “summarise”, and even “automate your workflow” — have long been sitting right in front of us, with or without LLM intervention. But even more than that, clichés and tropes cannot be separated from a distinctly human register of time, as well as a distinctly human mode of sociality. This is true whether — as in this novel — we’re talking about a team of neurodivergent colleagues who don’t need much technical training to spot discrepancies in data, or about a tech mogul in the making who comes across research at the right time and right place, research that wouldn’t exist had it not been for years of academic fieldwork, and a path to that fieldwork already paved by senior scholars. Bix Buxton’s Own Your Unconscious does not spring fully formed from the mind of one “genius”, but instead draws (steals?) from anthropologist Miranda Klein’s famous study called Patterns of Affinity: a “slender monograph” containing “algorithms that explained trust and influence among members of a Brazilian tribe”. It’s in fact because of Klein’s algorithms that Bix’s company amasses “a fortune”. This might seem like a fictional liberty, until you consider that Peter Thiel’s foundation funds the René Girard scholarship, that Steve Jobs was inspired by Ayn Rand, and that Alex Karp likes to reference his former professor Jürgen Habermas.10 Indeed, all of this might seem far-fetched until you consider how much of the technological “innovation” of genAI is stolen labour, from peer-reviewed academic articles to the whole of what we’ve been doing online, without knowing that what we were really doing was making Meta’s Llama models immanent.

In a recent ChatGPT commercial, we’re shown into a trendy apartment where a conventionally attractive young heterosexual couple gets ready to have a meal together, Perfume Genius playing.11 The man watches eagerly as his date takes a bite. “Good?” he asks. “Really good,” she says, nodding as she chews. The background music gets louder as they settle in for the rest of their date. The camera pans out, and a scrolling text of a ChatGPT log comes into the foreground. The prompt reads: “I need a recipe that says ‘I like you, but I really want to play it cool.’” In response, the chatbot generates “here’s the move”, proceeding with a recipe and justification. The message ends with a reassurance: “above all, don’t sweat it. You got this.” What’s there to worry about? ChatGPT can code this stockblock: Nervous Protagonist Tries to Impress Date with Home-Cooked Meal. If there’s a stockblock, “AI” knows just “the move”.

“Here’s the move” has become one of the most recognisable clichés of GPT-speak. It suggests that the “move” has already been discovered and that the act of discovery matters little to the endgame. OpenAI did a series of these elegantly produced 30-second ads where ChatGPT is consulted to solve familiar quandaries: how friends can plan a road trip, how an entrepreneur can start a new business, how a college student can better study for an exam, and how a new driver can fix an issue with a car. That last one isn’t like the others, but OpenAI would like you to think that it is. Fixing a car involves a clear answer. The other quandaries are steeped in contextual variables, requiring imagination, creative vision, and flexibility. “Here’s the move” encapsulates how Big Tech seeks to alter our habits of character and what we subsequently learn to value: they would have us believe that there’s no fundamental difference between fixing a car and building intimacy with a romantic partner. Either can be manufactured: the process, the way you get there, doesn’t matter.

That’s also what Bumble founder and CEO Whitney Wolfe Herd would like you to believe. In March 2026, Wolfe Herd announced that Bumble was going to be “saying goodbye to the swipe and hello to something that I believe is revolutionary for the category”. And nothing says “revolutionary” these days like integrating “AI”: reports indicate that Bumble plans to introduce Bee, a “personal AI assistant [that] would study a user’s interests, preferences and behaviour patterns before interacting with thousands of other AI systems to identify the most compatible matches”.12 As with OpenAI’s commercial messaging, Bumble’s initiative presupposes that there’s an answer to how and why people connect romantically.

The question of the process of chemistry, or why two people end up hitting it off, is central to the chapter “Rhyme Scheme”, where Lincoln — a neurodivergent “senior empiricist and metrics expert” — pursues his office crush M by running algorithmic models to determine the likelihood that she’ll fall in love with him. The only thing the models can’t predict or identify is the variable x, that “ineffable, unpredictable detail that makes one person fall in love with another person.” Obsessing over x, Lincoln sees things around him like a “blue stuffed pillow at Walmart”, and wonders if that might be x: the “unknown value required to secure M’s love”. “Secure” — there’s that “knowing everything” once again. There’s that algorithmic certainty. We might read x as the “move” that OpenAI offers, as if the messy riddle of a relationship comes down to one answer, one “x”.

When OpenAI’s GPT-3 launched in 2020, researchers were keen to see why so much hype and excitement were coalescing around this particular machine learning model for natural language processing. Philosopher Shannon Vallor was one such researcher; her 2021 article “The Thoughts the Civilized Keep” provides a revealing window into initial receptions of GPT-3. Vallor’s background is significant here: her first book retheorises virtue ethics to meet the conditions of digitality, asking if there are “habits of character ideally suited to our new technosocial circumstances?”.13 Virtues, Vallor tells us, are the ethical dispositions formed from actions done not just habitually, but over time and in alignment with broader, pro-social values. From Aristotelianism, Confucianism, and Buddhism, Vallor makes explicit that virtues and the habits they emerge from are contingent on shifting variables. The medium is, after all, socially the message: technological interventions gradually require adjustments to practices of, for instance, honesty or civility. All of this is important because to virtue ethicists, virtues are a primary way to think through how a specific group can flourish, together.

And so it makes sense that what Vallor saw with GPT-3 was not more yummy profit for shareholders, nor an enchanting display of competence, but the absence of something that she has long considered essential to humans’ flourishing: understanding, once again. In Vallor’s words, understanding is “not an act, but a labor” of history, consensus reality, and the social contracts that coordinate us toward something beyond violence or brute self-interest.14 Understanding is a “sustained, social project”, one that we “pursue daily as we build, repair and strengthen the ever-shifting bonds of sense that anchor our thoughts to the countless beings, things, times and places that constitute a world.”15 It is this labour, and not next token prediction that Vallor associates with thinking.

By the end of “Rhyme Scheme”, Lincoln has given up on trying to algorithmically win M’s love. She attends a party that Lincoln throws at his house, and when they go into his room she sees the collection of all the objects he has bought, in the hopes that they might be x. He tells her that they are just random items. They look out of his bedroom window, and their breath is taken away by the stars, by “all that math, glittering and shimmering back at us”. They share a moment. There was no “move”, although, as Lincoln proceeds to narrate to us, that very window was his x for buying the house, so they may have been affected by the same x, after all. Who’s to say when it’s “all just retroactive math”?

Big Tech’s “here’s the move” paradigm “is partly a language problem”, Rebecca Solnit reminds us.16 Silicon Valley recruits us to adopt its products through language: any time someone says the word “agent” in reference to a chatbot, for example, you have to wonder if they’ve ever really considered the fraught question of autonomy. Subway slop and the “techspeak spillage” of billboards full of sweet B2B nothings are all in the service of this recruitment.17 These recruitment strategies matter because language has always been more than mere mediating symbols. Language is a building block for values, and Big Tech’s values do not include Vallor’s labour of understanding — which encompasses, in Solnit’s words, the “arduous, the uncomfortable, the slow and wandering, the unpredictable, the vulnerable or risky, the intimate, the embodied”.18 With this in mind, Solnit encourages us to discover and develop languages, commitments, and dispositions that reclaim the sociality of building meaning and interpretation.

In the centre of this tension, The Candy House affirms and maintains our joint human meaning-making project by dramatising where this social labour of understanding foments in human communication: within clichés, of all things. While its 19th-century legacy as an “overly copied figurative phrase to avoid” has stuck,19 rhetorical and popular engagements with the cliché and its constellations — tropes, stereotypes, aphorisms, maxims, dead metaphors — demonstrate just how special it is that we have these instantly persuasive placeholders built and refined over time and rhetorical ecologies. Clichés and their ilk are so special, in fact, that writers can’t help but create analogies when articulating their very functions: thus, Rivka Galchen compares clichés to “old talismans dug up at an archaeological site”, and Leslie Jamison describes them as “subterranean passageways”.20 They quicken dialogue, foster interaction, and make room for multiple agendas. Their familiarity is an invitation to participate: horror film tropes prime you to not to get too invested in the first character(s), and therapy-speak TikToks package canned wisdom that hits like perspective. Clichés are social to their core.

The final chapter in The Candy House explicitly models the affordances of such social and socialised human clichés — resources that will never be “the move” even as they are infinitely more compelling, more fitting, and sometimes even more efficient than the GPT date-night meal or the Bumble match, ironically enough. Throughout The Candy House, we keep hearing about this forgotten middle child, Ames, whose two brothers have already taken centre stage in earlier chapters. Finally, though, it’s Ames’ turn. Egan takes us back to his childhood, to a little league baseball game in the early 90s. The unremarkable Ames is at bat.

Incredibly, against all odds, Ames hits a game-winning home run, instantly making him the stuff of local legend. Theoretically, this could forever change the trajectory of his life. That would be “the move”. Through meta-commentary, Egan acknowledges that it is “tempting to build a version of Ames’s life in which [his little league home run] was a ‘turning point,’ empowerment at a crucial moment, etc. (stockblock 3Miis), but that would be phony”. Phony, and all too statistically probable as a literary move — just as later, when Egan paints a picture of Ames having a conventional married-with-children life. No sooner does she set that scene than she takes it apart, pointing out that there is a “false bottom under the happy ending, a hollowness around the words”; asking, “can you hear it?”. For Ames, there was “no wife … [there] were no children except the ones Ames imagined he heard when he sat in his father’s old study”. At every turn, we expect growth and closure, but Egan refuses, insisting that there are “gaps”, even with the assistance of a technology like Own Your Unconscious. It is fiction and only fiction that lets us “roam with absolute freedom through the human collective”; so we end the chapter where we started, with everyone celebrating Ames’ big swing: the authentic, immeasurable, and banal highlight of a life.

This ending is decidedly not “the move”. Had the novel ended with Bix, the tech mogul that it begins with, everything might have tied together neatly, with meaningful catharsis and parallelism. Instead, Egan chooses the anti-climactic ending: not a cell phone in sight. Watching Ames at bat, his mother’s emotions are cliché to “anyone who’s read a book or seen a movie about children playing sports and how their mothers feel”, and yet the scene still feels singular — partly because of where it falls in the novel, sure, but more so because clichés are gathering places for individual data points. It turns out that the statistic entity can’t be separated from individuals and their lived experiences. The Candy House reminds us that human storytelling matters now more than ever because writing has always been a technology of human sociality, reliant as it is on recurrent infrastructure that we the people build.

In 2018, Sam Altman put down a $10,000 deposit, to be put on the startup Nectome’s waiting list to upload his consciousness when he dies, arguing in 2017 that “superhuman AI is going to happen, genetic enhancement is going to happen, and brain-machine interfaces are going to happen. It is a failure of human imagination and human arrogance to assume that we will never build things smarter than ourselves.”21 This invocation of “human imagination” is rather interesting. While “artificial general intelligence” or “artificial superintelligence” remains a eugenics-fueled conspiracy theory, Big Tech CEOs also promote ways of living in the here and now that are antithetical to the value of imagination implicit in Vallor’s social labour of understanding.22 Satya Nadella, for example, says he enjoys podcasts; but instead of listening to them, he “loads transcripts into the Copilot app on his iPhone so he can chat with the voice assistant about the content of an episode”.23 Nadella also boasts that he doesn’t read work-related emails, choosing instead to rely on Copilot to “deliver summaries of messages”. Big Tech companies have hoovered up countless resources, exploited and traumatised innumerable contingent labourers, and burned through more than a trillion dollars, and the best that they can come up with is an eschatology devoted to efficient email and more job precarity — “no-holds-barred capitalism”, in Ted Chiang’s words.24

We are reminded of a curious concept from The Candy House: that of Anti-Vision. When Bix’s children are older, he shows them his memory of the night he came up with Own Your Unconscious to instil in them the magic of human imagination. The memory starts with what Bix calls “the Anti-Vision: a vacancy where a new idea refused to appear. For several moments it filled the screen, depthless and white.” His son, Gregory, is fascinated, wondering if that vacancy only looked empty. Gregory himself has refused to upload his consciousness, unlike his friends and most around him. The things that matter to Gregory — “books and writing” — mean that he grows apart from his father and his world of “technology, wealth, fame”. He gets his MFA but stops writing when Bix is diagnosed with ALS. Still, Gregory can’t help narrativising the lives of the New Yorkers he can see from his apartment. When it dawns on him that he really must finish his book, he’s walking back home on a snowy day and finds himself staring “into a gray-white void”. In it, he recognises his father’s Anti-Vision, but where Bix saw a frustrating lack, Gregory understands that the Anti-Vision “had never been an absence—the opposite! It was a density of whirling particles. His father just hadn’t gotten close enough.” That density was a “galaxy of human lives” that only from a distance “faded into uniformity”. Up close, though, and with the requisite patience, curiosity, training, and, indeed, imagination, Gregory witnesses the collection in the collective.

Both of us have taught The Candy House at the college level, and this time around some students wanted Egan to be more critical of Big Tech, but details like the Anti-Vision deliver a biting and prescient critique: Altman and his cohort are irrelevant. Their attempts to standardise humanity by rendering humanity’s shared labour of understanding obsolete will only ever leave them on the outside. They don’t get that “things are more connected than they seem”. To them, being “smart” can only mean more knowing — more “moves” and fewer movements to get there. They lack the capacity to appreciate the machinery of literature, language, and “old talismans”, so they fall back onto capitalism and bad readings of Snow Crash and The Terminator. How embarrassing.

We should be clear: details like the Anti-Vision deliver insight only through close reading, re-reading, and conversations with other people. Without that slow, shared labour, it’s likely that you’ll miss what makes this novel essential. Indeed, critical responses to The Candy House registered disappointment that The Candy House seems to lack the sort of personal intimacy that is a hallmark of Goon Squad. Thus, critic Lauren Oyler writes that: “Egan’s relentless genre bending threatens the narrative immersion that is her greatest strength.”25 But Oyler’s commentary actually speaks to a central strength and strategy of this novel. Egan took the machinery of fiction — all the affordances of section breaks and arrangement, multiple networks and media, and your standard typographical choices like whitespace and italics — and re-mediated context collapse into literature.26 The remediation of the logics of algorithmic feeds into a book asks questions about how immersed we can really be even as its scope and scale immerse us in connection. What a marvel that we can still do anything and connect with anyone in the middle of all this noise. The Candy House is deliberately choppier than Goon Squad because it is formally demonstrating context collapse and its effects on not just our abilities to tell stories full of both evocative clichés and stockblocks, but also on our abilities to interpret and make meaning with stories. Perhaps critics like Oyler needed to talk about this novel with other people. Perhaps she needed to take more time.

Maybe it’s time we all slow down.

In a recent Claude ad, a young woman lies on the floor reading a book.27 Time flashes forward: she’s reading Penguin classics, then Foucault, then Kant — we can hardly keep up. She’s sitting up now, laptop perched on her legs, and she tells us that she’s been thinking about time. “We can’t necessarily change how we perceive time,” she observes with studied eloquence, “but we can resist how it rushes us”. To practice this resistance, she continues, she’s been using Claude to help her think through a reread of Kant’s Critique of Pure Reason. Where she might have normally just moved on from a difficult passage, now she slows down and thinks it through with Claude. When she wonders what Augustine might have said in response to Kant, she turns to Claude. “It’s not about speed, it’s about sitting with questions,” she concludes — a bit of that familiar “AI” “it’s not X, it’s Y”, the appropriation of our rhetorical device of negative parallelism. She doesn’t use Claude to answer questions, she insists, but “as a tool to slow ideas down”.

The repeated use of negative parallelism shows the extent to which Claude is advertising itself in response to critical engagements contra “AI”. Anthropic is positioning Claude as a new modality of thinking: “Maybe if we start thinking like this, time won’t always seem like it’s running out,” our Claude-partnered Kant scholar offers as a final techno-utopian salvo. Like much “AI”-speak, it’s superficially compelling until we consider the unspoken assumptions and ideologies. First of all, if anyone made us feel like time was running out, it’s Big Tech’s hyper-distribution of capitalism, so let us be clear that they are selling us a solution to a problem that they helped cause. But more importantly, why shouldn’t time be a limited resource? Time will always run out. “Time’s a goon,” as The Candy House’s predecessor said; and yet, Big Tech still aims to fight it, whether with Altman wanting to upload his consciousness, or with Peter Thiel’s commitments to transhumanism and cryogenics.

The ad is in line with Anthropic’s push to brand Claude as a “thinking partner” and “the AI for problem solvers”. While ChatGPT is for the masses, Claude is for the discriminating few — and you’re one of them. Anthropic’s ad is tailored to those of you reading this article, telling you that our gen-AI isn’t ruining creative thinking; it’s amplifying it. It’s a tempting candy house. Critical thinkers, it says to us, are in fact the ones best equipped to “harness the power of AI”. Isn’t “deeper” thinking what we’re after? And if only we had more time, couldn’t we then think more deeply?

Arne De Boever, in his recent article The Essay vs. AI: On the Literary Value of Criticism, offers a more satisfying alternative. What “AI” is doing, De Boever argues, is not altogether new. He argues that criticism has already “been in the process of becoming formulaic, and in our time AI is accelerating this process, an acceleration that is further contributing to the demise of the humanities”.28 Just as Egan shows that fiction has always been reliant on tropes and clichés, so, too, does De Boever suggest that “AI” is holding a mirror to what he sees as the self-standardisation of criticism. For De Boever, this is a call to action: “If AI can appear to us, then, as the miserable outcome of this long institutional history of criticism’s practice, my position is that we are being solicited, today, in the time of AI, to write otherwise.” It’s fundamental that criticism — literary and otherwise — pursues different routes, dwells on different texts, and models different writing than what has become increasingly standard.

We aren’t suggesting a reactionary zagging just because “AI” is zigging. The way “AI” zigs, however, might be information about how, when, and why we might zag in response as a disenfranchised people during a period of technofascism. If, as De Boever suggests, we’ve been overdue for a zag, let us take cues from The Candy House’s renewed emphasis on what Ruha Benjamin calls “ancestral intelligence”: or the “insights, experiences, and wisdom that grow under the rubble of progress”.29

It’s no surprise that Egan uses the redemption narrative as a sort of foil in The Candy House. After all, it’s one of the easiest stockblocks for a writer to pull off. Where Goon Squad is full of this stockblock, The Candy House, however seduced, refuses to give in. When a former opioid addict turned beloved small-town politician narrates his turnaround, he remarks how “improbable” it must sound, but he “invites us to recall the narrative power of redemption stories”. Yet a redemption story is too much like knowing everything — a cliché too well-worn to mean what it used to. It has, like the writing “AI” produces, become too much “the move”. We might do better to recall our ancestral intelligence: the unparalleled joy of a random home run, the idiosyncratic recipe cobbled together from texts to your family’s group chat, the reading group where hours are spent bumbling through philosophical obscurities. The Candy House asks us to reconsider what inherited knowledge we want to keep, but also what we want to refuse, which are questions we can similarly ask ourselves as writers, thinkers, educators, and critics: what tropes of language, structure, and thinking have delimited us or mean to delimit us? What practices, scenes, and forms of collaboration may help us zag otherwise?

Notes

  1. José van Dijck, “Tracing Twitter: The Rise of a Microblogging Platform”, International Journal of Media and Cultural Politics, 2011, p. 340. [^]

  2. Caddie Alford and Jonathan S. Carter, “‘A Cesspool of Toxicity, Hatred, and Discrimination’: Twitter, Free Speech Absolutism, and Adoxastic Enshittification”, Journal of Contemporary Rhetoric, 2025. [^]

  3. Roger Ebert, X, 29 August 2011. https://x.com/ebertchicago/status/107973508637790209. [^]

  4. Jennifer Egan, “Black Box”, The New Yorker, 28 May 2012. [^]

  5. Deborah Treisman, “This Week in Fiction: Jennifer Egan”, The New Yorker, 25 May 2012. [^]

  6. The DISCO Network, Technoskepticism: Between Possibility and Refusal, 2025, p. 8. [^]

  7. If you’re curious why we use scare quotes to refer to “AI”, a good place to start would be Emily Tucker’s “Artifice and Intelligence”, Tech Policy Press, 16 March 2022. [^]

  8. Matt Seybold, “‘You Have to Use it. You Have to Trust it.’: Forced Adoption of AI is the Subtext of Davos”, The American Vandal, 23 January 2026. [^]

  9. Jennifer Egan, A Visit From the Goon Squad, 2010. [^]

  10. Adrian Daub, What Tech Calls Thinking: An Inquiry into the Intellectual Bedrock of Silicon Valley, 2020. [^]

  11. OpenAI, “Dish with ChatGPT”, YouTube, 28 September 2025. [^]

  12. Swastika Sruti, “Could AI Soon Control Your Love Life? Bumble’s New Dating Vision Sparks Panic and Curiosity”, Oneindia, 14 May 2026. [^]

  13. Shannon Vallor, Technology and the Virtues: A Philosophical Guide to a Future Worth Wanting, Oxford, 2016. [^]

  14. Shannon Vallor, “The Thoughts the Civilized Keep”, Noema, 2 February 2021. [^]

  15. Vallor, “Thoughts”. [^]

  16. Rebecca Solnit, “What Technology Takes from Us—and How to Take it Back,” The Guardian, 29 January 2026. [^]

  17. Clio Chang, “The Year of Subway Slop”, Curbed, 29 December 2025. [^]

  18. Solnit, “Technology”, 2026. [^]

  19. Ruth Amossy makes this clear in her affirmative theory of clichés vis-a-vis reading in “The Cliché in the Reading Process”, Substance, 1982. [^]

  20. Rivka Galchen and Leslie Jamison, “Why Do We Hate Cliché?”, The New York Times, 6 January 2015. [^]

  21. Antonio Regalado, “A Startup is Pitching a Mind-Uploading Service that is “100 Percent Fatal”’, MIT Technology Review, 13 March 2018; Sam Altman, “Words that Matter”, Medium, 7 December 2017. [^]

  22. Will Douglas Heaven, “How AGI Became the Most Consequential Conspiracy Theory of Our Time”, MIT Technology Review, 4 November 2025. [^]

  23. Austin Carr and Dina Bass, “Microsoft’s CEO on How AI Will Remake Every Company, Including His”, Bloomberg, 15 May 2025. [^]

  24. Ted Chiang, “Silicon Valley is Turning into Its Own Worst Fear”, BuzzFeed News, 18 December 2017. [^]

  25. Lauren Oyler, “Future Nostalgia: Jennifer Egan’s Old-World Internet Novel”, Harper’s Magazine, April 2022. [^]

  26. The term “context collapse” comes from: Alice E. Marwick and danah boyd, “I Tweet Honestly, I Tweet Passionately: Twitter Users, Context Collapse, and the Imagined Audience”, New Media & Society, 2010. [^]

  27. @ontaphilosophy and @claudai, “How would you rebel against time?”, Instagram, 19 February 2026. https://www.instagram.com/reels/DU8UX-_jJQr. [^]

  28. Arne De Boever, “The Essay vs. AI: On the Literary Value of Criticism”, b2o, 2026. [^]

  29. Ruha Benjamin, “The New Artificial Intelligentsia”, Los Angeles Review of Books, 18 October 2024. [^]