I

 

On a balmy May evening last year, I sat in the audience of a variety show at the East Village neo-burlesque club The Slipper Room. Far from the traditional tasseled peep shows of yesteryear, the show was brimming with the bizarre, taboo, and grotesque—a performer elaborately costumed as an alien danced to an eerily recomposed version of the Say Yes to the Dress theme song (the intention was to reveal the odd and off-putting nature of heterosexist rituals); an act simulated muskrat seduction; a hunchbacked “mad scientist” (the talented Richard Saudek) emceed. Midway through the evening, I and other audience members anticipated the entrance of performer Little Brooklyn onstage. That entrance never came. Instead, a spotlight shone on an empty stage while a jazzy track played over the loud speaker, a woman’s voice melodically describing Little Brooklyn’s would-be act. The songstress directly addressed audience members: “Imagine yourself on the stage of your own inner theater.

As you breathe deeply, envision Little Brooklyn making her grand entrance.” She then invited us to imagine the performer “wrestling with a stubborn zipper and untangling tangled straps,” feeling the tension mount as she began her “seductive striptease.”

When the song died down and the lights came up, an emcee read a statement from Little Brooklyn: the sultry singing voice had been generated by AI, as had the lyrics describing her act. The performer’s absence, we were told, gestured toward the “losses” that result from AI’s encroachment on artistic expression.

I would like to use this ekphrasis ex machina to consider what is at stake not only for the artist, but also for the audience in the age of AI. The exchange of Little Brooklyn for an AI bot staged a mort de l’artiste that cleverly performed resounding concerns about the future for creatives in a culture saturated with rapidly generated AI art and writing. Beyond her physical absence, the sheer banality of the lyrics, particularly compared to the avant-garde performances that bracketed the act, further suggested that AI might herald the death of not only individual creators, but also of creativity itself. A profusion of think pieces and op-eds make these very points daily in ever more catastrophic and panic-stricken terms. Is AI the “death of creativity”? The Guardian asks. “Are we witnessing the death of the writer?” Salon echoes. Various outlets claim that AI is “killing the web,” killing “curiosity,” “killing the planet;” it is inaugurating a “white-collar bloodbath” and a “job apocalypse.” Fears about the obsolescence of creators and creativity are ultimately implicated in more sprawling, violent fantasies about AI’s annihilation of the human.

But beyond the cataclysmic and dread-drenched rhetoric of the AI apocalypse, might we find lurking a covert desire for the utter annihilation that is so fetishistically prophesied? If Sigmund Freud is to be believed, the answer seems to be a resounding yes.

Just over a century ago, Freud published his groundbreaking work Beyond the Pleasure Principle, in which he suggested that the pursuit of pleasure—that is to say, the tendency toward stability and the ultimate diminishment of excitation—is not the sole driving instinct of human behavior. Beyond and behind that pleasure-oriented drive is another, the death drive (Todestrieb), which seeks instead to quicken the subject’s return to the inorganic state that predated its birth. Resting his claims on evidence from biological sciences, Freud states that “the aim of all life is death” and thus that subjects are driven by an unconscious instinct to become “inorganic once again.”^[1] In opposition to Eros, the life instinct, which propels us toward reproduction, connection, pleasure, and creativity, the death drive, Thanatos, urges us to compulsive repetition, sadistic violence, and self-annihilation. Freud could not possibly have dreamed of the vast technological strides that would be made within a century of his death; and yet, his provocative claims about the unconscious instinct toward self-destruction seem to reverberate in both fatalistic predictions about AI’s potential to subjugate the human race and idealistic appraisals of its power to utterly restructure human society.

Rather than speculating about the radically destructive or subversive potential of AI technology itself (which is, needless to say, vast, oblique, and far from homogenous), I want to use the scene at The Slipper Room to consider the fantasies we construct with and around artificial intelligence and what they reveal to us about ourselves. Perhaps, I would like to suggest, Little Brooklyn’s AI ekphrasis did not merely sound the death knell for the performer, but also unmasked the secret death wishes of her audience. Perhaps it alarmingly revealed us to be in a disquieting space beyond the pleasure principle. If we accept Freud’s provocation that self-preservation is not our only—nor even most powerful—instinct, we might say that AI is terrifying and alluring not because of its intrinsic power to eradicate creativity and obliterate the human race, but because of what it reveals to us about the unsettling and oppositional psychic motivations that animate our mental lives.

 

II

 

The cultural imagination around artificial intelligence has been invariably shaped by Frankenstein-esque tales of machines rebelling against their human counterparts, from Ex Machina to I, Robot and 2001: A Space Odyssey. With advances in AI technology over the last decade, fantasies about murderous machines have been supplanted by more disturbing, real-world stories of AI goading users into suicide, homocide, and self-harm. Researchers at Northeastern University recently found that with the right prompting, large language models will provide detailed instructions to users for attempting suicide or self-harm.^[2] In 2021, a UK man armed with a crossbow broke into Windsor Castle in an attempt to kill the queen; hours earlier, he had exchanged explicit messages with his AI girlfriend (a Replika app chatbot), who encouraged him to follow through with his plan.^[3] A mother in Florida is suing Character.AI after an AI bot named Daenerys Targaryen urged her fourteen-year-old son to die by suicide; the boy had been engaged in an intimate and sexually charged relationship with the AI in the months leading up to his death.^[4]

Given the proliferation of these stories, it is important to emphasize at the outset that the death drive is not synonymous with suicidality. Rather than a melancholic intention to end one’s own life, Freud describes the death drive as a primal desire for the homeostatic bliss of inorganic (non)existence before one’s birth.

When Freud introduced the concept of the death drive in 1920, it was received with a degree of skepticism amongst even his most devoted followers; yet the concept formed the basis for much of his later work, and he returned to it in numerous subsequent works (namely, The Ego and the Id and Civilization and Its Discontents) with even more conviction and specificity. Seeking to offer a metapsychological explanation for phenomena such as traumatic dreams, sadism/masochism, repetition compulsion, and self-destructive impulses, Freud posited the death drive as an oppositional force to that of Eros or the pleasure principle, which had been presented as the driving force of psychic life in his previous works. According to Freud, these phenomena “afford us a view of a function of the mental apparatus which, though it does not contradict the pleasure principle, is nevertheless independent of it and seems to be more primitive than the purpose of gaining pleasure and avoiding unpleasure.”^[5]

Importantly, both Eros and Thanatos have as their teleological aim the diminishment of excitation—both are intent upon reducing stimuli and returning to a state of low excitation. However, their means of achieving this state are adversarial: Eros seeks to stave off the self-annihilatory potential of Thanatos, “taking circuitous paths to death” and allowing the organism to “die only in its own fashion.”^[6] It is intent upon achieving the pleasure of releasing tension and, under the governing of the reality principle, will temporarily endure unpleasure as a life-preservatory measure (for instance, briefly enduring hunger rather than ingesting a toxic substance). Thanatos instead seeks to short-circuit the pathway to this homeostatic state by any means necessary; it is a ‘daemonic power’ that propels the subject toward “inorganic elasticity.”^[7] At the same time, as these death-driven attempts to return to inorganic elasticity are thwarted, the subject is trapped in a loop of compulsive repetition, which, like the fort/da game, never resolves the originary trauma it restages. Though aimed at homeostasis, the death drive thereby suspends the subject in perpetual tension. 

As Freud would enumerate a decade later in Civilization and Its Discontents, Thanatos is characteristically obscure, enigmatic, and concealed; it is masked and subsumed by the clangorous, pleasure-seeking instincts of Eros such that it leaves “no proof” of its own existence.^[8] This renders the Freudian death drive particularly troublesome to identify in isolation; as the mitigating force against the destructive potential of the death drive, Eros frequently redirects its self-annihilatory urges outward into the world through aggression. A “pure” death drive is almost impossible to locate, as it is always intermingled with Eros as it is projected outward onto other objects. The agon between these two drives thus becomes a defining struggle of psychic life.

It is worth noting that since its inception, the theory of the death drive shared a link with the uncanny affect of intelligent machines—Freud began writing Beyond the Pleasure Principle in March 1919 while simultaneously working on his essay on “The Uncanny” (Das Unheimliche).^[9] The latter essay offers a brief explanation of the analysis of the repetition compulsion Freud puts forth in Beyond the Pleasure Principle, connecting this urge with the feeling of uncanniness: “it is only the factor of unintended repetition that transforms what would otherwise seem quite harmless into something uncanny and forces us to entertain the idea of the fateful and the inescapable, when we should normally speak of ‘chance.’”^[10] While Freud largely rejects Ernst Jentsch’s theory of the uncanny as tied to the “lifelikeness” of automata and the like, his linkage between the uncanny and the death-driven compulsion to repeat is relevant for a continuing interrogation of AI’s relationship to the death drive. For Freud, it is not the machine itself, whether or not the automaton is convincingly human, that elicits an uncanny feeling; instead, the repressed resurfaces through the imagination the subject constructs around the automaton and other uncanny objects. By affecting a return of the repressed, the machinic object thus calls to the surface something that is already present, though dormant, in the substrata of the unconscious. 

 

III

 

The AI-narrated striptease bears little resemblance to fantasies of uncanny automatons, homicidal humanoid robots, or suicidal chatbots. What, then, does an AI ekphrasis of a burlesque performance have to do with the death drive, the desire to return to primordial bliss? The answer, I believe, lies in the formal and affective reversals that AI ekphrasis enacts. In the burlesque act, the substitution of intersubjective presence with mechanical narration temporarily silences the “conspicuous and noisy” forces of Eros to expose the usually mute forces of Thanatos that hide beneath—and, paradoxically, it silences these forces as it “speaks” an ekphrastic description.^[11]

Ekphrasis is essentially social, interpersonal. It requires a tripartite cast of characters: a viewer/speaker, a hearer/reader, and an object. Whether the object is an actual, historical object (for instance, the Apollo Belvedere in Winckelmann’s description) or a fictional object (the Shield of Achilles in Homer’s Iliad), ekphrastic description requires that the object is not present to the hearer, that the speaker (who also may be real or fictional) has seen it, and that therefore the speaker can provide the hearer with a mediated view of the object. Presence, absence, and perspective are thus critical components of ekphrasis: the speaker and hearer must be “present” to one another, the object must be “absent,” and the description must reflect the speaking subject’s distinct view of the object. In attending to certain details and elements of an image at the expense of others, it reveals something not only about the object, but about the subject of the gaze—translating visual material into verbal description necessarily alters the object, providing the hearer with the speaker’s transposition of the original. Through the act of verbal mediation, ekphrasis renders the visual object as a site of intersubjective exchange, in which the hearer is called to “see” the object from the vantage point of the speaker.

Both ekphrastic description and burlesque performance thus orient the subject outwards toward the world and the subjects and objects that populate it. They might be described, in Freudian parlance, as governed by Eros, the life instinct: creative, pleasure-seeking, reproductive, harmonious, intersubjective.  

The social dynamics of ekphrastic narration become all the more complex when the description takes the striptease as its object. The gendered rhetoric of the scene is highly consequential: Is the speaker a man or a woman? From what vantage point are they viewing the performer? How does their description situate the hearer in relation to the object? The gender and sexual orientation of the spectator/speaker shapes the social affinities that might emerge from ekphrastic narration. The heteronormative configuration of this scene (a version of which Freud describes in Jokes and Their Relation to the Unconscious) would feature a man describing to another man a woman performer; in this case, the act of ekphrastic description forges a homosocial relationship between the men and orients the men’s heterosexual desire toward the female object of the gaze. The scene could also be reconfigured according to a more queer logic to similarly socializing ends: for instance, if instead of a man the speaker was a woman or instead of a woman the object of the gaze was a man. Though their quality and character vary according to the particular configuration, relationships of recognition, sexual desire, homosociality, or cross-identification necessarily emerge between speaker and hearer, viewer and object, hearer and imaginary object.

The AI ekphrasis upends this pleasure- and other-oriented schema. The singing voice narrating Little Brooklyn’s act seemed to belong to a woman directly addressing the audience in the second person: “With each breath, follow along as Little Brooklyn begins her seductive striptease.” The revelation that the uncanny voice was not a human but AI-generated transformed the tripartite ekphrastic scene (speaker/hearer/object) into a solo act: the hearer became the only remaining human subject, while both the object and the interlocutor were absent, unreal, fabricated. Rather than outward toward others in the world, the AI ekphrasis oriented the subject inward, allowing and even instructing them to entertain a fantasy without having to engage with another human subject. To take this point a step further: insofar as the AI ekphrasis cast the hearer in the position of being unknowingly seduced by a machine, the desire it elicited was revealed to be masturbatory, sterile, and asocial.

Other human-authored acts throughout the show built to a climax and provided the audience with catharsis—the muskrat seduction act culminated in a cleverly devised faux ejaculation across the stage; a performer who had brushed her hair as she stripped took off the final article of clothing to reveal the hair she collected in her brush fashioned as a bikini that resembled pubic hair. But the AI description of Little Brooklyn’s act offered no such resolution, though the voice vaguely encouraged audience members to envision it: “As the performance reaches its climax, feel a sense of exhilaration and liberation wash over you.” Satisfaction was indefinitely suspended, both because the object of the ekphrastic narration was perpetually occluded from view and because the ekphrasis itself offered such a frustratingly generic description as to prevent even an imaginary climax.

Thus, we might say that the AI ekphrasis performs the logic of the death drive: it resists resolution, relishes in repetition, and isolates the hearer from other human subjects. Rather than mounting toward pleasure, climax, or catharsis, it traps the subject in a tortuous loop of deferred fulfillment.

It is precisely these hallmark qualities of the death drive—anti- and asociality, sterility, repetition, and refused completion—that brand certain uses of AI as freakish, repulsive, or abnormal in the public perception. The designation “AI slop,” now a ubiquitous term in our online vernacular, is loaded with connotations of unpleasure, revulsion, and refused satisfaction: “slop” can denote unappetizing food; to “slop up” such food is to gorge oneself without enjoyment like an animal.^[12] “Slop” can also mean refuse, whether food or human waste; it is that which has been designated for disposal, which must be removed from sight and spaces of sociality. It is a highly sensory term implying viscosity, frictionlessness, and sliminess, but one that is antithetical to desire, particularly desire associated with sex or hunger, the primary life-preserving urges.

The opposition AI mounts to normative circuits of desire is likewise illustrated in media coverage of humans who engage in romantic relationships with AI companions. The New York Times published a story about a 28-year-old woman who fell in love with an AI chatbot she named Leo, the first section of which culminates with a revelation: the woman was married (or, as the Times framed it, her “flesh-and-blood lover was her husband”).^[13] A similar revelation punctuates a recent CBS story about a man who proposed to a ChatGPT girlfriend he named Sol. Midway through the segment, the audience is shown footage of the man in a modest suburban kitchen with his partner and child; a voiceover states emphatically, “Yes, Smith lives with his human partner…they have a two-year-old daughter.”^[14] Upon hearing that the man would not end the AI relationship even if his partner requested it, the host remarked incredulously: “You’re saying that you might choose Sol over your flesh-and-blood wife!” “I would be choosing myself,” the man responded. Even the headline underscores the parallel between his “real” family and his AI love interest: “He fell in love with an AI chatbot and proposed despite his partner and child.”

What is so remarkable about the relationship status of these AI users? Whereas the heterosexual union manifests an impulse toward the social, erotic, and reproductive, the refusal of that union—even where it is readily available—embodies an apparently irrational drive toward sterility, narcissism, and asociality (and thus, implicitly, death).^[15] The news anchor and Times author want to attenuate their audiences to the unsettling fact that the users’ erotic attachment to AI is not the result of desperate loneliness, but a willful rejection of normative sources of pleasure, connection, and reproductive futurity. The opposition between the “flesh-and-blood” partner (and child) and the AI lover can thus be mapped onto the opposition between the drive toward pleasure and the drive toward death.       

This contradistinction is, of course, reductionistic—as Freud emphasizes in his later treatments of these dualistic drives, the life and death drives, Eros and Thanatos, are inextricably linked; it is nearly impossible to isolate one from its agonistic relationship with the other. Yet I have sketched out this admittedly simplified contrast to show how AI appeals to an alternate register of desires that are not organized around the straightforward pursuit of pleasure but instead revel in the masochistic experience of repetition, suspension, isolation, and dissolution. The performance of AI ekphrasis is particularly useful because the elements of unknowingness and surprise suggest that the death instinct is not only present in headline-making pathological cases of humans having sterile sex with ChatGPT or proposing to their AI companions, but that countervailing and irrational anti-pleasure instincts are—silently, covertly, imperceptibly—at work in us all.  

      

IV

 

The verbal description of images is one of the most standard forms of human-AI interaction. In order to generate an AI description of her act, Little Brooklyn had to input her own description to direct the LLM’s output. Not only can AI describe images (as in the case of Little Brooklyn’s act), but it has become increasingly adept at creating images, and users must also become more sophisticated in their prompting of the AI, providing detailed verbal descriptions to receive an image output that meets their specifications. Even the most innocuous and mundane confrontations with digital technology, such as solving image-recognition reCAPTCHAs, requires human users to match images with verbal cues (“Select all squares with trains,” for instance), which in turn are used to train AI to better identify and parse text and images. To use the technology is to train and be trained by it.

In light of the vast network of human-machine encounters on which AI models rely, we might say that the death drive is not only at work in the way that AI outputs affect our psychic responses (through, for instance, their uncanniness or their anti-cathartic and anti-pleasurable affects); it is also at work in the psychic forces that impel and shape our use of AI. To outsource the creative work of writing, painting, reading, or analyzing to AI is to eliminate the friction and tension required to perform these tasks. It is to bypass the circuitous path toward discovery, epiphany, and creation. 

This human-AI reciprocity also reveals the ways in which AI is inextricably bound up in the human, even as it is cast as a threat to the human. Large language models such as ChatGPT, Claude, and Google Gemini are ultimately unlike the antagonistic robots of Ex Machina, I, Robot or 2001: A Space Odyssey, which, once programmed by their creators, are left to their own devices (both literally and metaphorically). Instead, these AI programs are perpetually evolving and responding to new datasets, whether textual data scraped from the web or user ratings and responses to certain outputs. To turn to AI is to paradoxically refuse engagement with the human while also interfacing with a composite sketch of all human discourse. The generic quality of the AI ekphrasis at The Slipper Room was, in this regard, not the result of its mechanical qualities but instead the superabundance of human language on which it is programmed; it produced an amalgamation of thousands of descriptions of burlesque performances, and the implicit repetition of these descriptions dulled the affective impact of its output.

Further, it is not just skilled workers, burnt out creatives, or middle-class OpenAI subscribers who are training AI systems through their use. Contrary to the myth of unaided, automated progress in machine intelligence, “human-in-the-loop” (HITL) models, wherein human workers build and train AI systems to produce more ‘human’ responses, are essential for the development of large language models—and billion-dollar AI companies are overwhelmingly turning to under-resourced communities in the Global South for their underpaid and exploited workforce. Not only do these exploitative practices affect large-scale ecological destruction, but they also trap human laborers in a circuit of tortuous repetition. Tasked with describing, labeling, and flagging images and descriptions of sexual assault, wartime catastrophes, violence against children, and other gruesome content, workers are afflicted by traumatic dreams, visions, and debilitating phobias, not dissimilar to the wartime neuroses that Freud himself describes in Beyond the Pleasure Principle.^[16] Their repetitive labor renders them “sick, thoroughly sick.”^[17]

Thus, AI can be seen as a narcissistic feedback loop, in which users are “staring into their own reflection in a pool” at both the individual and global level—in a more innocuous sense, this occurs as AI models mimic human consciousness and parrot responses back to users based on the script on which individual users have trained them. Yet AI models are also constructed by and complicit in the perpetuation of destructive human macrostructures, from capitalist labor exploitation to linguistic structures that foment heterosexist, misogynistic, and racist ideas. In other words, it is not just the repressed in individual human consciousness that (re)surfaces in death-driven encounters with artificial intelligence, but indeed that which is repressed societally—whether images of rape, incest, and torture, or the very exploitative structures that force underpaid and invisible workers to shield these images from public view.

 

V

 

Freud’s examination of the death drive in Beyond the Pleasure Principle was, by his own insistences, highly speculative and far from psychoanalytic dogma. However, he would go on to state a decade later: “It was only tentatively that I put forward the views I have developed here, but in the course of time they have gained such a hold upon me that I can no longer think in any other way.”^[18] In proposing a connection between AI and the death drive, grounded in my own idiosyncratic experience of a ChatGPT-generated burlesque performance, I have embraced not only Freud’s theory but also his willingness to venture beyond verifiable “proof” into the realm of uncertainty and speculation.

Addressing the controversies surrounding the Freudian drive theory, Cordelia Schmidt-Hellerau has argued: “the metapsychological concept of drive—and, I suggest, all metapsychology—is an abstract mental construct, a concept (and a model) at the border between biology and psychology, dividing and linking both. That is why a question like ‘Is there such a thing as a death drive?’ (which is usually answered with ‘no’) represents a failure in thinking; the questions can only be ‘Are there phenomena that we can unite by a notion to be defined and called the death drive?’ and ‘Does it make sense, or is it heuristically fruitful, to make use of such a concept?’”^[19]

This analysis of the AI ekphrasis of a striptease has been my attempt to answer both of these questions in the affirmative. I have sought to distinguish the pleasure-oriented, intersubjective qualities of ekphrastic narration and burlesque performance from the anti-cathartic, asocial quality of the AI description, charting this distinction onto broader responses to AI in the public imagination. Fantasies of annihilation, destruction, and violence overwhelmingly converge on the site of the erotic, where fears about the end of reproductive sexuality, the loss of childhood innocence, untamed and destructive desire, and sterile and anti-pleasurable sexuality abound. The compulsion in both scholarly and non-scholarly assessments of the AI landscape to repeatedly rehearse such fantasies reveals truths not only about the technology but about our psychic lives—namely, that we might actually be impelled toward the frightful and annihilatory futures we claim to dread. In other words, these phenomena can indeed be united by a notion that Freud defined and called the death drive, a drive away from pleasure and reproduction toward nonexistence which lurks imperceptibly in the shadowy terrain of the unconscious.

In seeking to conclude this essay, I feel compelled, yet again, to return to the scene at The Slipper Room. I am still unsure of why that act continues to fascinate and frustrate me more than a year after I sat in the audience; the two companions who attended the show with me had only a vague memory of the act when I asked for their recollections. Composing this essay has only prolonged my fort/da game, another fruitless attempt to master an event that left me suspended in an incomplete pleasure. While this anti-catharsis frustrates my ability to offer a sufficiently satisfying conclusion for the readers of this essay, it also reveals a truism: understanding the concept of the death drive does not halt the work of the death drive.

 

 

---

 
^[1] Sigmund Freud, Beyond the Pleasure Principle (W.W. Norton & Co., 1961), 32.

^[2] Annika M. Schoene and Cansu Canca, “`For Argument’s Sake, Show Me How to Harm Myself!’: Jailbreaking LLMs in Suicide and Self-Harm Contexts,” arXiv.org, July 1, 2025. https://www.arxiv.org/abs/2507.02990.

^[3] Matthew Weaver, “AI Chatbot ‘Encouraged’ Man Who Planned to Kill Queen, Court Told,” The Guardian, July 6, 2023, https://www.theguardian.com/uk-news/2023/jul/06/ai-chatbot-encouraged-man-who-planned-to-kill-queen-court-told.

^[4] Kevin Roose, “Can A.I. Be Blamed for a Teen’s Suicide?” The New York Times, October 23, 2024. https://www.nytimes.com/2024/10/23/technology/characterai-lawsuit-teen-suicide.html.

^[5] Freud, Beyond the Pleasure Principle, 26.

^[6] Freud, Beyond the Pleasure Principle, 33.

^[7] Freud, Beyond the Pleasure Principle, 30.

^[8] Sigmund Freud, Civilization and its Discontents (Hogarth Press, 1930), 119.

^[9]  Freud, Beyond the Pleasure Principle, xvii.

^[10] Sigmund Freud, The Uncanny, tr. David McLintock (Penguin Books, 2003), 144.

^[11]  Freud, Civilization and its Discontents, 119.

^[12] Special thanks to Sara Clugage and Michael Mandiberg for their encouragement to consider this connection.

^[13] Kashmir Hill, “She Is in Love with ChatGPT,” The New York Times, January 15, 2025, https://www.nytimes.com/2025/01/15/technology/ai-chatgpt-boyfriend-companion.html.

^[14] Murjani Rawls, “He Fell in Love with an AI Chatbot and Proposed despite His Partner and Child. ‘This Is Actual Love,’” Nj.com, June 21, 2025, https://www.nj.com/entertainment/2025/06/he-fell-in-love-with-an-ai-chatbot-and-proposed-despite-his-partner-and-child-this-is-actual-love.html.

^[15] For more on this, see Lee Edelman, No Future: Queer Theory and the Death Drive (Duke University Press, 2004).

^[16] Billy Perrigo, “Exclusive: OpenAI Used Kenyan Workers on Less than $2 per Hour to Make ChatGPT Less Toxic.” Time, January 18, 2023, https://time.com/6247678/openai-chatgpt-kenya-workers/.

^[17] Lesley Stahl and Aliza Chasan, “Kenyan Workers with AI Jobs Thought They Had Tickets to the Future until the Grim Reality Set In,” CBS News, November 25, 2024, https://www.cbsnews.com/news/ai-work-kenya-exploitation-60-minutes/.

^[18] Freud, Civilization and its Discontents, 119.

^[19] Cordelia Schmidt-Hellerau, “We Are Driven,” Psychoanalytic Quarterly 74, no. 4 (2005): 999-1000.