Simulacra in the Age of AI: Baudrillard and the Hyperreality of Generated Signs

From Lascaux II to the Matrix. AI art

Introduction

The rapid proliferation of artificial intelligence has transformed how texts, images, and even voices are produced. These outputs often have no direct counterpart in the world—they are generated patterns, assembled from prior data but not tied to any original source. How should we interpret a photograph of a person who never lived, or an essay created without authorial intent? Jean Baudrillard’s Simulacra and Simulation (1981/1994) offers a striking conceptual framework. His theory of simulation, the four phases of the image, and the emergence of hyperreality illuminate not only the logic of media culture in general but also the particular dynamics of AI-generated content.

The Fate of the Image

At the outset of The Precession of Simulacra, Baudrillard outlines four successive stages in the destiny of the image:

1. Reflection: The image reflects a profound reality—a portrait or religious icon that purports to mirror something external and sacred.
2. Distortion: The image masks and alters reality, as in propaganda or advertising that reframes perception.
3. Concealment: The image conceals the absence of a profound reality, sustaining the illusion of depth where nothing lies beneath. Lifestyle campaigns promising “happiness” through consumer goods exemplify this.
4. Pure simulacrum: The image has no relation to any reality whatsoever. “It is its own pure simulacrum” (Baudrillard, 1994, p. 11).

This schema sets the stage for a radical claim: signs do not merely distort or conceal reality but ultimately generate a world without origin. Baudrillard stresses that “simulation is no longer that of a territory, a referential being, or a substance. It is the generation by models of a real without origin or reality: a hyperreal.” Once images reach this final phase, they circulate autonomously, producing their own truth-effects.

Models Without Territories: From Borges’s Map to AI

Baudrillard begins with Borges’s fable of cartographers who create a map so detailed it covers the territory itself. Traditionally read as an allegory of representation, Baudrillard inverts it: today it is not the map that follows the territory but the territory that decays beneath the persistence of the map. What remains are only “the shreds of the real” scattered across the desert of simulation.

This allegory anticipates AI in uncanny ways. Large language models and generative systems do not correspond to the world directly. They are “maps” built from textual or visual data, models of models, detached from any originating reality. Yet these maps do not merely describe reality—they precede it, shaping what counts as meaningful or plausible. When ChatGPT or Stable Diffusion generates output, it is not reproducing a world but conjuring one through patterns of signs. The model engenders the territory, not the other way around. AI thus literalizes Baudrillard’s “precession of simulacra.”

Simulation, Simulacra, Hyperreality

To understand this logic, Baudrillard distinguishes between representation and simulation. Representation presupposes a stable correspondence between signs and reality, even if imperfect. Simulation dissolves that bond: the sign is no longer an image of the real but a system that produces reality itself.

A simulacrum is precisely such a sign: a copy without an original, a model that fabricates rather than mirrors. Disneyland is Baudrillard’s classic case. It is not simply a fantasy land but a deterrence machine. It presents itself as imaginary to persuade us that the “outside” world is real, when in fact all of Los Angeles and the America surrounding it are already of the order of simulation.

In the digital present, Meta’s metaverse performs the same function. By advertising itself as an artificial, bounded virtual space, it reassures us that our offline world is real. Yet our everyday lives—already saturated with social media feeds, algorithmic curation, virtual economies, and “authentic” influencers—belong to hyperreality. The metaverse, like Disneyland, conceals that the outside has already collapsed into simulation.

Replicas and Substitutes: The Lascaux Caves

Baudrillard recalls the decision to close the prehistoric caves of Lascaux to visitors to preserve them, replacing access with an exact replica constructed nearby. From that moment, the distinction between original and copy collapsed: the duplication sufficed to render both artificial.

This logic resonates with AI-generated content. The “original” is inaccessible or irrelevant; what matters is the circulation of convincing substitutes. A synthetic face produced by a generative model does not imitate a particular person—it is a Lascaux replica, a perfect substitute that abolishes the need for an original referent. In training data, too, AI feeds on endless reproductions, recycling signs until the origin is no longer necessary.

The Lascaux case dramatizes the AI condition: the real is no longer hidden behind appearances but has vanished altogether, leaving us with convincing replicas that are “truer than true.”

The Hyperreal Everyday: From the Louds to Influencers

Baudrillard also discusses the 1971 television experiment filming the Loud family, billed as an unscripted glimpse into “real life.” The family eventually disintegrated under the gaze of the camera, raising the paradox: did television record reality, or did it produce the very crisis it claimed to observe? The program offered not reality, but a frisson of the real—a simulation more captivating than reality itself.

This anticipates contemporary influencers and reality streaming, where carefully staged performances masquerade as “authentic” lives. AI avatars now extend this logic: synthetic influencers with no human behind them generate shows, interact with followers, and embody brands. Like the Louds, they offer the thrill of intimacy, but this intimacy is engineered, calculated, and ultimately hyperreal.

Actualization in the Digital Landscape

Baudrillard’s categories find concrete realization in contemporary media. Photography once reflected reality (stage one). With editing tools, it could distort (stage two). Social media influencers exemplify stage three: podcasts that present curated “authenticity” mask the absence of an unmediated self. Today, AI-generated images fully inhabit stage four. A synthetic portrait of a non-existent person is pure simulacra—a sign without referents.

AI intensifies the loop Baudrillard foresaw: signs produce signs. Models are trained not on reality but on corpora of texts and images—signs upon signs. Their outputs then re-enter circulation, shaping discourse and future models. AI is not a distortion of reality but a generator of hyperreality.

Relevance Today

Baudrillard’s framework clarifies three dimensions of our digital condition:

1. Epistemological stakes: In hyperreality, the line between truth and fiction collapses. “Post-truth” politics is less a breakdown of fact than the triumph of simulation.
2. Ethical stakes: If authorship and authenticity collapse into circulation, how do we evaluate responsibility? An AI-generated fake speech can have consequences regardless of its factuality.
3. Cultural stakes: Memes, cryptocurrencies, virtual economies, and the metaverse are not mirrors of reality but new symbolic orders. Like Disneyland or Lascaux II, they sustain themselves as replicas, not as representations.

Conclusion

Artificial intelligence, far from being a neutral technology, makes visible the logic Baudrillard diagnosed decades ago. It exemplifies the shift from representation to simulation, culminating in hyperreality, where signs proliferate without referents and in turn shape what counts as real. The Borges map, the Lascaux replica, the Louds’ televised life—all were early signals of a culture in which the model engenders the world. AI simply accelerates this dynamic, turning hyperreality into the infrastructure of everyday life.

As Baudrillard (1994) suggested, “there is no longer a difference between the real and its double” (p. 12). To understand our present media ecology, we must see AI not as an imitation of reality but as one of its primary engines. We do not live after Baudrillard’s theory—we live inside it.

References

• Baudrillard, J. (1994). Simulacra and simulation (S. F. Glaser, Trans.). University of Michigan Press. (Original work published 1981)
• Gunkel, D. J. (2007). Second thoughts: Toward a critique of the digital divide. Critical Studies in Media Communication, 24(2), 187–202.
• McIntyre, L. (2018). Post-truth. MIT Press.
• Paris, B., & Donovan, J. (2019). Deepfakes and cheap fakes: The manipulation of audio and visual evidence. Data & Society.
• Poster, M. (2001). What’s the matter with the Internet? University of Minnesota Press.
• Wardle, C., & Derakhshan, H. (2017). Information disorder: Toward an interdisciplinary framework for research and policy making. Council of Europe report.
• West, D. M. (2019). How to combat fake news and disinformation. Brookings Institution.