In the years since the November 2022 release of ChatGPT, “GenAI” (generative artificial intelligence) has become a cultural obsession, fueling the dreams and, importantly, financial investments of corporate executives, policymakers, military leaders, and university presidents. A collective sigh, meanwhile, emerged from scholars in the humanities and interpretive social sciences, to whom the new technology’s primary use-case seemed to be cheating on undergraduate essays. Yet, as Lauren Goodlad notes in her introduction to the inaugural issue of the journal Critical AI, “critical thinking of the kind that interpretive disciplines foster” is sorely needed at the current juncture. What kind of digital beast was being unleashed on the world in the mid-2020s? Was AI, as Silicon Valley experts alleged, a totally transformative technology that would end all white-collar work and transform human creativity? Or was that apocalyptic vision just clickbait and marketing hype? Maybe AI was just another enshittifying technology imposed by gadget-crazed tech barons on an already beleaguered humanity and climate-crisis-doomed earth? Or maybe it was a little bit of all of the above: transformative, or at least useful, in areas such as medical research; but also a massive water, land, and energy suck, as well as yet another tool to concentrate wealth in the hands of the few.
This forum, which grew out of an interdisciplinary panel at University of Toronto—the sometime-home of AI “Godfather” Geoffrey Hinton—started as a conversation about “alignment.” In tech-policy speak, this term of art denotes the artificially narrowed question of whether AI can be made to “align” with human values, or whether intelligent machines will one day appropriate the world to their own ends, with or without a moral compass. This question is often posed in a way that seems to (conveniently?) forget just how plural and contested “human values” are. It also, with its mythic projection of a coming robot God that is only semi-controllable, works to capture conversation about the risks of AI, diverting from other crucial topics like the consolidation of corporate power. Our conversation started with alignment, and the desire to seek it, as a “prompt” for exploring how the study of religion, in collaboration with fields from media studies to law to computer science, can contribute to the critical interrogation of AI. The discussion developed in all kinds of intelligently generative ways. In this forum, we encounter AI as deity or superintelligent bodhisattva (Anna Su). We discover its entanglements with evangelical Christianity (John Lardas Modern), proselytizing “New Atheism” (Suzanne van Geuns), and Hindu and Muslim content moderators in Bangladesh (Ishtiaque Ahmed). We see what happens when a media studies scholar not primarily interested in religion (Sarah Sharma) discovers the God rhetoric that keeps popping up in her published work.
At the time of writing, an ever-expanding pantheon of commercially branded robot helpers (ChatGPT, Claude, Perplexity, Midjourney, Copilot, etc.) are weaseling their way into the fabric of daily life in the guise of psychotherapists, lovers, oracles, teachers, copy editors, and personal assistants—consuming water and land and then belching out not-so-angelic clouds of carbon with every interaction. Meanwhile, heralds keep trumpeting the world-altering potential of these digital entities. We are living, as Modern observes, through a “season of revival” and “tremendous ferment” filled with sermonistic effusions, rampant myth-making, and apocalyptic eschatology that can feel an awful lot like religion—still more specifically, like the nineteenth-century Second Great Awakening that left parts of New York State feeling like a “Burned-Over District.” It bears repeating that today’s world is burning over for more literal reasons: the aforementioned greenhouse gasses, combined with AI datacenters’ ravenous need for fresh water and expanses of land, whether in a suburban warehouse district or a prairie grassland, including on sovereign Indigenous territories.
If AI feels like a tent revival, this is because—as Su explores in her contribution—it is but the latest chapter in the much longer history of what historian David Noble long ago dubbed “the religion of technology.” Despite their breathless presentation of their products as always already radically new, AI enthusiasts replicate key features of this earlier history—in their techno-utopianism, their passion for transcending bodily existence, and their obvious fervor for godlike entities. What else is “the Singularity,” that coming AI superintelligence, if not an apocalyptic fantasy about a messianic robot God?
This religio-technological mythos has a quite specific genealogy. Silicon Valley sprouted from California’s distinctive mélange of counter-cultural ideologies and spiritual and lifestyle experiments, from the Whole Earth Catalogue to the Esalen Institute. Largely imagined and enacted by white men who were at once engineers and seekers, this mythos continues to generate new sensational forms (borrowing from Birgit Meyer) for the twenty-first century, blurring the line between “secular” and “sacred.” That line, as the field of critical secular studies has demonstrated at length, is itself historically contingent, locally specific, and routinely renegotiated. Tech culture has its own spiritual idioms and texts, and its own semi-religious imperative to “Work Pray Code.” Occasionally, these religious connections come dramatically into view, as in Peter Thiel’s “Antichrist Lectures,” which combine the biblical book of Revelation, Carl Schmitt, and René Girard, whose Stanford lectures Thiel attended as an undergraduate.
The religious impulses embedded in the genesis and justification of AI pervade and animate Silicon Valley’s techno-utopian culture. Sometimes, this techno-religion does the classically ideological work of obfuscating power structures. It deflects from the very real material consequences of critical mineral extraction, cooling systems, and undersea cables by prompting us to imagine data floating ethereally in “the cloud.” It also allows the barons of digital capitalism to cement (while clouding) their status by presenting their power and privilege as eternal verities. Technologies do not drop from the sky as immutable gifts from the gods. They are developed in particular ways to meet particular social needs, and it always bears asking whether and how they could have been developed differently—as Raymond Williams demonstrated in his bracing critique of technological determinism, prompted by the age of television but equally relevant to AI. The technology “exists,” the tech-barons tell us, and resistance is futile—as though it cannot exist differently, be reshaped, or be questioned. Resistance may or may not be futile, but that is a social and political problem, not a technological problem.
Technological determinism has thus become a genius marketing tool of enchantment, with experts literally invoking magic to describe AI’s inner workings. Once again, the Internet is proving itself (in Wendy Chun’s phrase) “wonderfully creepy,” simultaneously enticing and alarming us with new—and potentially dangerous—marvels. Van Geuns, in her contribution to this forum, suggests that scholars of religion are especially well-equipped to bring “attention to the architecture of wonder, to the violent snips and spirals that produce a new age in real time.” She offers “AI secular religion” as “an anchoring term for where to look when humans assign miracles to a coming age.”
Is this age of digital wonders still a secular age? Or should Silicon Valley’s techno-religion prompt a routine “update” (as it were) of our theorization of secularism? Charles Taylor’s notion of the “immanent frame” revealed secularity’s unnatural naturalness by, among other things, tracing the historical sidelining of notions of divine intervention and the rise of self-contained or “buffered” subjects. With AI systems vacuuming up our language and interior lives, we humans might now seem more porous and less buffered than before—although dodgy and unequal internet access entails its own form of “buffering.” Disconnection is a defining feature of all networks, including messianic networks. Meanwhile, AI seems to promise if not divine intervention exactly, then at least digital intervention—a deus in machina that creates new openings between humans, machines, and the systems in which we live.
Far from heralding a radical break with the human, then, AI is a retrospective tool of word and image recombination that works to re-code human stuff: racism, sexism, geopolitical inequities, and more. Search engine algorithms have been shown to perpetuate racism, and AI seems poised to offer more of the same. By training themselves on massive datasets of the already written and spoken, Large Language Models (LLMs) come to reflect the most statistically dominant attitudes in their archive—which is to say, to recycle stereotypes of all kinds, including stereotyped clichéd language.
Like other forms of digital production, AI also relies on often-occluded human labor. In Ahmed’s contribution to this forum, we encounter a group of Bangladeshi “annotators” hired to label online videos as “toxic” or “safe” according to rubrics set by Global North bureaucrats—adding structural alienation to affective injury, as these workers themselves become repositories of affect, taking in the bad feelings that companies do not want to circulate to others. As always, history matters here. Digital products, largely owned by companies in the Global North, especially the United States, rely on and perpetuate neocolonial structures in the Global South, even as they deplete the lands and waters of their own settler-colonial territories. Tech companies’ censorship and “annotation” practices inherit and build on colonial laws that are still in effect across Bangladesh, India, and Pakistan, working to govern wounded religious feelings and, along the way, regulate the boundary-lines of religious communities.
The history of “AI” extends across the postwar period and across a series of efforts to, as Modern argues, build “neuromatic” machines, which would replicate the structures and functions of the human brain. Torturing chimpanzees and other primates was central to this research. For a long time, the gold-standard for AI was whether it could ace the so-called “Turing test,” passing itself off as convincingly human. Goodlad suggests that the “ELIZA effect,” named after Joseph Weisenbaum’s 1960s software program ELIZA and denoting humans’ remarkable tendency to anthropomorphize, might be the better touchstone. As David Hume put the matter in his 1757 Natural History of Religion, “There is a universal tendency among mankind to conceive all beings like themselves…. We find human faces in the moon, armies in the clouds.”
Both the Turing and ELIZA standards contain intensely gendered conceptions of possible and desirable relations among human, non-human, and machine beings. Popular memory of Alan Turing, as captured in the movie Imitation Game, posits that his adult desire for machine intelligence sprang from the void left by his first human romantic love—the boyhood friend to whom the closeted Turing, living in a Britain where homosexuality was still very much illegal, could not confess his desire. In the 1950s, Turing was convicted of engaging in homosexual acts and then, in period argot, “chemically castrated.” Meanwhile, Weisenbaum’s ELIZA program was named after Eliza Doolittle of My Fair Lady and George Bernard Shaw’s Pygmalion, which retold the Greek myth about a sculptor who falls in lust/love with a female statue that comes to pervy life, making it a modern, sadistic tale about a posh man educating a Cockney woman to speak correctly. Desire for computer companions, in other words, is thick with the violence of heteropatriarchal and classed desire, something that Donna Haraway pointed out many moons ago.
Sharma and Van Geuns both bring gender into clear view in their work on what they respectively call “insufferable tools” and “computational creep.” Van Geuns shows how the current impulse to regulate life through algorithms and metrics emerged in tandem with seduction coaches, incels, and other manosphere subcultures that take other beings (mostly women) as objects of computational manipulation. “In the manosphere, the improvement of social and even biological intelligence requires its artificialization,” she writes. Sharma takes us to the dark heart of such subcultures: “Mommy’s Basement.” As she persuasively argues, the technological ecosystem that has emerged from Silicon Valley over the past two decades has effectively replicated the garages, basements, bedrooms, and other suburban-domestic spaces where so many of these technologies were first incubated, resulting in apps and an attendant “gig economy which has been designed by our tech bros to fulfill all the labor one’s mommy/imagined mommy no longer does for them”—a world of “inexhaustible assistance.” Gendered labor imagined as willing care gets displaced onto classed and often racialized labor. “Is this really what we all have? Was this all there was to design?” Sharma asks.
Semiotic practices and technological systems encode a whole archaeology of gendered desire linked to desires for god-figures, as Sharma incisively shows. If the tech baron is a would-be god, flush with cash and know-how, his seeming sovereignty is threatened by his Pygmalion helper precisely in its/her maternal role. To reveal this implicit maternal dynamic, Sharma cites Hinton’s late-breaking AI regret: “If it’s not going to parent me, it’s going to replace me.” The tech inventor desires a created mother, not a Creator-mother— all-powerful but ultimately subordinate. He thus keeps on anxiously creating his own parent in a perversely Oedipal loop, haunted by a Freudian God.
Where does all of this leave us? What is to be done? We hope that, at minimum, approaching AI via the study of religion shows our tech-obsessed culture in a new light, revealing it to be stranger, more interesting, more filled with possibility than it sometimes seems. We also hope it leaves us wanting more and better questions to be asked of AI cultures—the kind of fulsome cultural analysis for which the humanities and interpretive social sciences are known. What does it mean when language finally comes to function as a machine, marching forward on its own through a logic of computation? Should we be working toward what Ahmed, in this forum, calls a “postsecular AI,” or is AI already postsecular, insofar as it shares in what van Geuns calls a “secular religion”? Do we need a new era of constitutionalism, as Su suggests, this time targeted at machines not kings? Such questions may or may not resolve themselves in the near future—if ever. For now, let us simply close on the sanguine note that Monique Scheer sounded in the conversations that led up to this forum. New technologies have prompted panics before. Most of them end up being neither everything nor nothing, but something in the mundane in-between. Though it may feel different this time around, that is the this-worldly space where we pitch this forum.
