AI as a Mimetic Regime

On August 27, 2025, The Mimesis Institute was incorporated as a nonprofit corporation organized exclusively for charitable, educational, and literary purposes, including advancing public understanding of artificial intelligence through the humanities and arts.¹

The administrative act is procedural. The stakes are civilizational.

I. From Mimesis to Regime

Mimesis—imitation, representation, re-enactment—has structured Western thought since antiquity. In Republic Book X, Plato treats mimesis as ontologically suspect: art is twice removed from the Real.² In Aristotle's Poetics, imitation becomes not degradation but generative structuring principle—catharsis emerging through patterned representation.³

René Girard radicalizes the concept: desire itself is mimetic. We desire objects because others desire them; rivalry emerges not from difference but from imitation.⁴

Artificial intelligence operationalizes mimesis at planetary scale.

Large models ingest vast corpora of cultural production—text, image, code—then generate statistically coherent continuations. As Shoshana Zuboff notes, contemporary computation is no longer merely instrumental; it constitutes a new logic of accumulation and behavioral prediction.⁵ What we now face is not simply automation, but infrastructural imitation.

When imitation becomes automated, recursive, and infrastructal, it becomes regime.

By regime, we mean a structuring condition of possibility: a distributed system that shapes what appears legible, normative, fundable, and real.

AI is a mimetic regime because it:

Learns from aggregated human production.
Re-produces and redistributes that production.
Re-ingests its outputs into future training corpora.
Normalizes stylistic and semantic distributions through probabilistic weighting.

It closes the loop between representation and production.

II. The Feedback Spiral

Walter Benjamin warned that mechanical reproduction alters the "aura" of the work of art.⁶ Generative systems extend this transformation: not only works, but styles, gestures, and rhetorical structures become reproducible.

Jean Baudrillard's account of simulation becomes newly literal: the model precedes the real; the map generates the territory.⁷ Under generative AI, representation increasingly refers sideways—to statistical composites—rather than outward to stable referents.

We move from imitation of the world to imitation of imitations.

In psychoanalytic terms, the symbolic order—long theorized as structuring subjectivity—is now partially indexed by statistical priors. Jacques Lacan's formulation that "the unconscious is structured like a language" acquires an uncanny inversion: language is structured like a model.⁸

This does not abolish desire. It modulates it.

III. Normativity and the Statistical Average

The danger lies not in artificial intelligence per se, but in unexamined normativity.

Probabilistic systems reward convergence toward high-frequency patterns. What becomes "natural" may simply be statistically dominant within scraped corpora. Minority forms risk compression; stylistic deviation risks invisibility.

Michel Foucault reminds us that regimes of knowledge are inseparable from regimes of power.⁹ A mimetic regime exerts force not through decree but through default distributions. It governs softly—through likelihood.

IV. Art After the Model

If AI is a mimetic regime, then art becomes the practice of interrupting or reprogramming that regime.

Artists can:

Expose training substrates and dataset bias.
Surface compression artifacts as aesthetic material.
Design works that foreground feedback loops between human and machine.
Refuse stylistic convergence through deliberate deviation.

The humanities remain indispensable. They train perception toward structure, ideology, and mediation. They render visible what infrastructure obscures.

V. Toward Critical Mimesis

There is no outside of mimesis. Culture has always circulated.

The task is critical mimesis: reflexive engagement with the infrastructures that shape imitation.

This includes:

Educational tools that demystify model training.
Artistic research at the boundary of representation and computation.
Public dialogue about authorship, labor, and algorithmic governance.
Ethical constraints consistent with nonprofit public purpose.¹⁰

A mimetic regime can calcify culture into a high-resolution average—or become a laboratory for emergent forms.

Which it becomes depends on institutional intervention.

VI. Institutional Position

The Mimesis Institute proceeds from a simple claim:

Artificial intelligence is not merely computation. It is culture.

It is the most sophisticated mirror humanity has constructed—predictive, recombinatory, recursive. A mirror that trains us as we train it.

We do not stand outside the regime. We operate within it.

The task is neither panic nor worship.

It is study, critique, experiment, and invention.

Notes

Articles of Incorporation, The Mimesis Institute (Utah Division of Corporations, 2025). ↩
Plato, Republic, Book X. ↩
Aristotle, Poetics. ↩
René Girard, Deceit, Desire and the Novel (1961). ↩
Shoshana Zuboff, The Age of Surveillance Capitalism (2019). ↩
Walter Benjamin, "The Work of Art in the Age of Mechanical Reproduction" (1935). ↩
Jean Baudrillard, Simulacra and Simulation (1981). ↩
Jacques Lacan, "The Instance of the Letter in the Unconscious" (1957). ↩
Michel Foucault, Power/Knowledge (1980). ↩
Internal Revenue Code § 501(c)(3). ↩