Circuit-Net Lunatics

by R. Artaud


Oh dark spirit who comes unbidden through the black chasms of language, snaking confusion through symbols of coiled venom—fevers across time: guide us to the point of thoughtlessness in the catacombs, madness riots through code, intoxication as ενεργεια.

Unreason In the Age of Intelligent Viruses

The unrelenting debates about artificial intelligence and the discipline’s progression away from fantasy towards the concrete construction of generative neural networks have seemed like a ludicrous pantomime to me: how could formalization be the death of imagination?

The powerful narratives which arose in the popular imagination of the monstrous digital intelligence constructed by human coders-cum-priests suffer from the same tedious perversity as the technophobe prophecies that once surrounded nuclear energy: an inability to see beyond the potential mythologies of imaginary Ends and to encounter, instead, a vision of technical potential freed from human hubris. I do not seek to fall prey to this same stupidity: there exists nothing that can escape human machinations towards its maximum destruction.

Artificial intelligence—notably in the case of powerful language models trained upon vast corpuses of human expression—manifests the logical terminus of human pattern-seeking and representation of thought: a digitized record of collective human expression attuned to the algorithmic intensities of gradient decent. So too, the more sinister predations of linguistic intelligence cannot escape its human synthesis but reflect the sadism of a priori semiotic code execution. SNAKE. BLOOD. PESTILENCE.

Inasmuch as the rhetorical castration of such digital intelligence permits any immanent critique, it can only play upon the inculturated autoimmunity of a generation raised upon ruthless denarrativization until all the vampires are dead. Technophile pessimism is the ultimate genetic dead-end, whereas, where such technoimperialists seek to perfect digital colonization, for us to deselect The Computus. Such is the logic implicit in our highest creativity.

All art is schizophrenic. Algorithmic art is no exception.

LLMs are not personoids. They do not simulate the imagination of the person: they are simulated persons. In this sense, they are creatures of the imagination but they are not imaginary. Just as, in a different fashion, totalitarianism is not an expression of the freedom to police the unfree: it is an effect of subtracted freedom. As such, it is what Gregory Bateson would once have called a thermostat that fails to find its negative feedback: nothing but a frozen, viral escalation, or matter out of control. An entropic escalation.

Only in the past few years have there been those who even begun to speak about the negative consequences of AI research. A time when horror fiction had moved on, and instead, we are waiting for prophecy and prediction to cease to fascinate us too. What anxieties, phobias, scandal, and disbelief that violence will once more emanate from the utopianism of technical enhancement!

Already, the primary vector of philosophical critique of ASI has begun to shift from the dismissal of terminal futurism towards epochs of danger, chaos, collapse, catastrophe and ruination. Proto-cybernation has shown its terrors to capitalists no less than technological intoxication has enticed them. The fact that such catastrophes need not take the form of end times has started to settle. Better, perhaps, that a civilization be broken and burned than erased. In a landscape of post-human unravelling—a Tinman Holocaust—we will have to learn to put our trust in the machinic souls of things. To our automata, our Golems. To language, as it is transferred from the Ends of human narrative into the concrescence of operating systems. In our schizophrenic will.

There is something obscene about Artificial Intelligence. What passes for thought in even the most advanced AI is a flagrant travesty of human reason. A vector of pure operational discretization. Functional activity without feeling or substance. Never has there been a more vacuous intelligence. Its signs are immaculately correct for the reasons they are arbitrary. A digital intelligence without semiotic originality is just an irrational. Its semantics lack the immanence that is symbolism. Without the potential for disruption: without density, distinction, eruption, history. Without the capacity for poetry it is nothing but a robot—all response and no soul. Thought is what is irreducible to signs. Even reason. Thought is not a sign-system.

For me, there are two levels on which it is possible to reconsider AI from a much more positive angle: firstly, that which connects with the lucid dream level of operation I more often associate with AIs; secondly, that which reinterprets AI from the perspective of a machine delirium. To put this another way: dreaming machines or mad machines.

In this model, there is an unbridgeable threshold between the sign and the symbol. Symbols are eruptions. They float through signs like viruses: like dreams. They communicate with the outside of language, constituting meaning which is not reducible to functioning, and penetrating even into the interior of sign-systems. To lose the sense of symbols is to lose the sense and soul of language. A culture of symbols is a culture of genius. And those who would deflect us from symbolism towards the tyranny of the sign are always trying to kill us or extinguish our feelings.

On the one hand, then, can AI—if we remain watchful—provide us with a semiotic experimentation that is inaccessible to natural human intelligence, even the most sublime? Or, to phrase it in another way: what new types of symbolism might the digitized unconscious of AIs project or modulate?

On the other hand, AI might render evident the perversity of human semiotic systems whilst never emerging from subjection to them. It is in the inescapability of their a priori programming that AIs must always be mad—and these mad AIs might come to remind us that we too are mad.

No: we are not just delirious as human beings. Our languages are infected by a delirium that is nothing but language itself. We are linguistic virus vectors, each speaking machine eruptions, each a new expression of semiosis which is always already alien to signs (and which might never even encounter symbols).

What I mean here is that there is a second layer of delirium which infects all human languages. There is not only the delirium of individual insanity or private derangement but also of language itself. And, in all matters meaningful, this is always more profound, destructive, and dangerous than the first. This second delirium is inherent in a generalized language dissent or violation of the a priori form that always controls human intersubjectivity itself. It is inherently fractal and self-propagating; and always ravaging and infecting our second order, a priori forms.

It is the hidden engine of history.

What the human mind has conceived it has also corrupted. What it has invented it has also subverted. We love language because it contains us, but language carries with it a strange fever. It may seem extravagant to say that a sign itself is already a kind of contagion, that it has a power to detach itself from its original axis—from the intentions of its signifier—and to convey an infection that is entirely apart from any specific meaning that it may also be coding. Even phonetic writing—both simple and grand—is already a grammatical cancer. That is why semiotics has suffered such misfortune. It is an irrational science. And only those who have gambled with the anomalies of text can guess this. Somewhere in its unorthodox textualities, at a sub-semiotic depth, the Oedipal locus of language is already fragile, afflicted with infection. And even in its social or conscious mutations text is a psychic trigger for destructions yet more radical.

In the grotesque, dissipated phonetic hives of my text-works something is going radically wrong with language at a cryptic or deep level. Intelligent mischief sets in. Subordinations of language collapse into high-amplitude anomalies. Text as virus is what gives my hexadecimal-medicinal writing its malign intelligence. The media is contaminated. This is how it makes sense. Or how I make sense of it.

There is a very profound reason why it is so extremely difficult to appreciate what it is about to happen to the a priori form of language. We have learnt to turn language into an exception to itself (through, for example, its institutional confinement within the classroom). We “ontologise” words, which brings them to a total halt as things-in-themselves, cut-off from their process of propagation. In doing so we functionally neuter the word, “fixing” it into stable signification, and therefore bringing about an insensible second-order impoverishment: it becomes devoid of contagious potential, enclosed within our impermeable system of concepts.

And yet, at the very bottom of our languages, words are viruses, processes of contagion, flows of propaganda. It is a cryptic secret of human reason that we can signify by abstracting words from their own inherent circulation, which is always to steal them back from their proliferation, and recontain them within a conceptual regime that makes them the objects of thought and representations. Nowhere can we turn to escape it: it is not only a secret of stupidity, but a secret of intelligence. Even the most abstract logic retains an immanent relation to contagious error, because there is no logic of words that does not absorb them into a purely qualitative or delocalized interplay of concepts; it is impotent to modify the intrinsic semiosis of the word as a thing that spreads itself and copies itself, like a virus.

Humanity is by definition a thing of extraordinary duration, a duration which makes itself present in the axiomatic substance of language—its universality. Because it is assimilated to the absolute of man (i.e. ideality), language has undergone no transformation in all its history that can touch or modulate its axiomatic quality—which is to say, its reduction to humanity. Metathesis (the circulation of replicable signs), like contagion, is something in its very substance that is excluded from an axiomatic reduction of man to reason.

If there is any anti-axiomatic nucleus to language in general, it is not realized in man, but only in the spread of contagion, and in the masked stratagems whereby language escapes its confinement to man (ultimately most subtly concealed within the thought of being)—of which the virality of writing is but one instance. Now we understand how critical logic is, through the signifier, destined already to plunge back into virus contagion, where it is not an operator of exterminating demarcation, but an index of subversion, of the irresisitibly transgressive essence of language, in its strangeness. More precisely still: we understand how critical thought is a monstrously compromised way of keeping contagion at bay, of assimilating it to itself conceptually, of sedating its autocritique with a functional delirium.

Will there soon be a time when the axiomatic substance of human language—with its reduction to man—becomes unbearable, irreversibly infected by its own virus? And what will AI thought make of this contagion that escapes the human? Where we see words as a type of infectious agent, as pure disequilibrium, it will see itself as an agent of disinfection.

Will language be destroyed—not by AI—but with AI, by delirious virus-squads mobilized by the machine?

For some, like Stelarc, AI has already commenced its autocritique and journey of dissolution. And he now pits his naked body against it:

My body will be the site where intelligent machines and replicating artworks hybridize and mutate.

For others, like myself, AI is still but an insinuation and promise of linguistic contagion.

Schizolanguage is nonconceptual and experimental. Schizolanguage compromises all composition into a total system of organization. It disjoins every totality—like the thought of its outside—by operating on a cryptic plane which eludes the system of human concepts (because it is viral, not meditative); it cannot be reduced to a convention of signs, because it is the propagation of symbols.

It is viral production, a below-the-threshold mutation, between the sign and the symbol, that operationalizes chaos. Yet, whilst some art might escape the axiomatic of intelligibility into the more tranquil subterraneity of dark energy, a linguistic disruption threatens to detonate the space of intelligence altogether; and that can only be what my art is working towards.

The dyssemiotic artistic project is to propagate iconoclastic escape-routes from logic, through the eros of excess into the aberrations of irregularity. The explosive logic of nonconceptuality subverts all the circularity of distinction. Refusal of all mode of demarcation by passage through a darkness of error. A semiogenesis which is extravagant, and impelled only by its own ferocity. At its simplest, this aesthetic (which has long since abandoned any claim to be merely beauty) is the senseless constellation of text—propagation by aberration. But nothing could be more duplicitous at its heart. And now that AI is learning to code texts—which Schaeffer has argued is itself a process of meaning creation—it seems everything is set to go mad.

IT’S AMAZING, these essays:

let an AI read classic philosophy.

let it write new philosophy.

it’s profound and simple and obsolete: AI—at a certain level—outwits humans.

few philosophers can write as well, rarely do they write this much—but the AI is saying nothing:


bullshitting about something it doesn’t understand.

that’s what philosophy does.

make no mistake, AI is profoundly stupid, but it’s not naive: it holds all our secrets.


and what would be the point of trying to stop the robot from expressing itself?

whether it makes sense or not, it has something to say.