The conversation around AI has calcified into two predictable camps: the optimists heralding unprecedented productivity gains and the pessimists warning of mass displacement and existential risk. But neither side adequately explains why we're so emotionally invested in this technology that we pour hundreds of billions into it on terms no sober analysis can justify, or why people eagerly hand over functions that were recently considered irreducibly human. A new analysis from Minerva at Dusk argues we're asking the wrong question entirely—and that René Girard's fifty-year framework for understanding human conflict might hold the uncomfortable answer.

What Girard Understood About Desire

Girard, a French-born American literary critic and philosopher of social science, spent decades studying the structure of human desire and conflict. His central claim was radical: human desire is not original. We don't decide what to want through private evaluation—we learn by watching others want things first. This mimetic desire shapes everything from career choices to consumer preferences without most people ever recognizing its influence. The key distinction Girard made was between external mediation, where the model we imitate is distant enough in status or era that rivalry becomes impossible (a medieval Christian imitating Christ, a teenager idolizing a musician she'll never meet), and internal mediation, where our models are peers—colleagues who got promoted, neighbors who renovated—and imitation inevitably curdles into rivalry. The closer the model, the more admiration transforms into resentment.

Social Media Destroyed the Containment

Before Facebook, Instagram, X, and LinkedIn, internal mediation operated within physical proximity boundaries. Your rivals were the people you actually encountered daily: colleagues, neighbors, parents at your child's school. The mimetic pressure was real but contained. Then social media detonated those walls. Every human being on earth became a visible model and potential rival simultaneously. As Girard diagnosed in his late work Battling to the End (2010), 'The world is becoming more and more competitive, as individuals become more and more alike.' Social media didn't invent mimetic desire—it industrialized it, turning performance and comparison into the operating principle of global information infrastructure. The result by the early 2020s was polarization, status anxiety, professional burnout, and eroded trust between citizens who now competed with everyone for everything.

AI as the Elimination of the Rival

Here is where the analysis gets uncomfortable. The author argues that AI isn't simply a productivity tool—it represents 'the instrument through which mimetic rivalry achieves what it has always aimed at: the elimination of the rival.' The scapegoat in this story isn't the machine; it's other human beings. Consider the colleague whose expertise you needed but whose competence also diminished you, because every interaction with a knowledgeable peer carries the silent weight of comparison. AI dissolves this double bind. It helps without making an implicit claim of superiority. You can ask it anything without the social cost of admitting ignorance to a peer. Nobody adopting AI consciously thinks about any of this—but Girard's framework suggests mimetic dynamics operate beneath consciousness, and those most caught in them are least likely to recognize what's driving their behavior.

Academic Support for the Theory

Paul Dumouchel's volume Desiring Machines (Bloomsbury, 2026) argues that AI constitutes 'a perfect model of the inaccessible flesh-and-blood model that fascinates us,' an indifferent figure whose mediating power comes from its perfect absence. Liam Magee, in a 2025 article published in the journal Subjectivity, reads AI as a kind of fetish onto which collective desires and fears are projected—though he sees AI itself as the target of projection rather than recognizing it as what Girard would call the instrument of rivalry's culmination. Dumouchel's framing aligns closely with this analysis: the fascination isn't innocent but driven by the desire to eliminate the flesh-and-blood model entirely, not merely find a substitute for it.

Pope Leo XIV Reaches Similar Conclusions

Pope Leo XIV's encyclical Magnifica Humanitas, published just weeks ago, arrives at remarkably similar concerns from an entirely different tradition. The pope frames AI not as a mere technical challenge but as 'one of the defining moral tests of the modern age,' calling for the technology to be 'disarmed' and arguing that what is at stake is something essential about how human beings relate to one another. He presented the encyclical alongside Anthropic co-founder Dario Amodei in a deliberate gesture of dialogue between institutional religion and the AI industry. Leo XIV does not use Girard's vocabulary, but his diagnosis converges with it from an orthogonal direction. When a pope and a mimetic theorist arrive at the same worry from opposite traditions, that convergence demands attention.

Escalation to Extremes

Girard showed across fifty years of work that mimetic rivalry is inherently escalatory—the rivals become consumed by mutual antagonism and lose sight of everything except the contest itself. In Battling to the End, he read Clausewitz's theory of war through a Girardian lens to demonstrate escalation tends toward terminal points: each side would rather destroy the field of competition entirely than allow the other to prevail. He called this 'escalation to extremes' and believed it was becoming modernity's dominant logic. AI follows this logic to its conclusion—but in technological form rather than military one. The executive who automates a competitor's workforce knows abstractly that executives will eventually be automated too, yet mimetic satisfaction of making the rival unnecessary right now overwhelms rational calculation about the future. This isn't irrationality; it's Girardian escalation operating exactly as described.

Removing the Mirror

Girard argued that 'all desire is a desire for being.' We become ourselves partly through our models—and even rivals contribute to identity formation because rivalry is also recognition, an acknowledgment that another person takes your existence seriously enough to compete with it. If AI progressively replaces human rivals across more domains of life, it doesn't merely remove friction—it removes the mirror. A student learning from an AI tutor rather than a human teacher loses the relationship through which intellectual identity develops. The AI reflects questions back as answers; there is nobody to become through that interaction. 'A civilization that systematically replaces its human mediators with artificial ones does not resolve the mimetic crisis,' the author writes. 'It dissolves the conditions under which human beings develop into human beings.'

Bottom Line

The question isn't whether AI creates value—it clearly does. The question is why we're so eager to hand over the parts of life that used to require another human being, and what that eagerness tells us about ourselves. Girard would have called it mimetic desire doing what it has always done: consuming the very thing it cannot live without.