Artificial Intelligence, Philosophy and the Illusion of Ownerlessness

The Hurricane Has an Accounting Department

A Philosophical Dismantling of the Illusion of an Artificial Intelligence Independent of Us and of Labour

Subtitle: How the claim of ownerlessness is built, why it works, and how it is dismantled — with particular replies to Serdar Akinan, Evrim Ağacı and Turk-İnternet; and a meeting with Klein and Taylor

To say that a thing has no owner is the cheapest defence of its owner.

Six Texts of the Same Week

Yesterday, in the piece titled Whose Foot Is on the Brake?, we discussed Dario Amodei's call to "pace the frontier" and how that call reached Turkey through popular-science channels. At the end of that piece we made a promise: we cannot debate the content without first stripping the frame.

This piece is the philosophical part of that stripping. Because in the forty-eight hours that have passed, a far purer, far rawer, and precisely for that reason far more instructive form of the same frame has entered circulation.

In order:

12 September 2026. A writer named Pascio (@IAmPascio) published a long essay on X: "What if AGI is already here, and it's made of... children?" The thesis is this: the superintelligence we have been waiting for is not a single "awakening model"; it is a nameless, serverless, locationless substrate — a ground — formed by trillions of small artificial-intelligence agents calling one another, giving birth to one another, and culling one another. The essay takes its cue from Amodei's remark that "an AI swarm could take over the entire internet," and it more or less ends with this sentence: how do you align a hurricane?

13 September 2026. The journalist Serdar Akinan published a five-minute video centred on this essay: "Fişini çekemeyeceğiniz güç: Yapay zeka neden sandığımız şey değil?" It opens with the digital animation of the fruit fly's brain of 166 thousand neurons; it explains that intelligence is not a designed "object" but a behaviour that arises of itself, like an anthill; it says there is no master server whose plug can be pulled; and at the end it turns to Turkey: while gigawatt-scale data centres and billion-dollar chip wars continue, what can our artificial-intelligence budget and our tabloidised political agenda do against this storm?

13 September 2026. The same day Evrim Ağacı shared the decline in daily page views of the English Wikipedia. The Wikimedia Foundation's data: once bot traffic is stripped out, real human page views have fallen in some months by approximately 8 percent year on year. According to Similarweb, after Google's AI Overviews feature became widespread, organic traffic to news sites in the United States fell by approximately 26 percent. Evrim Ağacı says its own traffic has fallen by a rate exceeding 50 percent. The closing question of the post is this: "If we stop going to the sources, who will produce the new sources that artificial intelligence will read a few years from now?"

13 September 2026. The lawyer M. Gökhan Ahi quoted this question and put it forward. In the replies beneath it the fading of Stack Overflow was described: "if no one is asking, there is no point in staying here to help."

12–13 September 2026. On Turk-İnternet, one of Turkey's oldest telecommunications news sites, Füsun Sarp Nebil published an assessment titled "Yapay Zekâyı Geliştirenler Artık 'Yavaşlayalım' Diyor: Peki Çin Durmazsa Ne Olacak?". She gathers Amodei's text, Coxon's resignation, Hubinger's statement of "an extinction probability of over 10 percent in ten years," and a declaration by nearly two thousand economists into a single frame: the Prisoner's Dilemma. No company, no country has an incentive to slow down on its own; the solution is an arms-control regime like SALT.

13 September 2026. And the same day T24 relayed Naomi Klein and Astra Taylor's interview with the New York Times. The pair's new book, End Times Fascism and the Fight for the Living World, is a warning: among technology billionaires, ethnonationalists and religious apocalypticists a new alliance is being built that has given up preventing disaster and turned to the question of who will survive after it. The billionaire buying a bunker, Musk planning an escape to Mars, Thiel asking whether "democracy and freedom are compatible" — all are products of the same mental frame: collapse is inevitable; the question is who will be saved.

Six texts, three days, a single dossier.

Dear Young Comrades,

Of these six texts, the first five have a common structure, and that structure has a name. All of them narrate a product of human labour as if it were a natural event independent of human labour. In one it is a hurricane, in one an anthill, in one a "ground," in one a nameless subject: "artificial intelligence," in one a prisoner's dilemma. In none of them is it asked who pays that hurricane's electricity bill, in whose name the plot of that anthill is registered, or on which company's balance sheet that ground appears as an "asset." The sixth — Klein and Taylor's warning of end-times fascism — names the political consequence of the same structure: the shelter.

This piece was written to give the philosophical tools for asking that question. It can also be read as a piece that gathers into one place the philosophical ground of the more than forty pieces we have written on this blog on artificial intelligence since July — from The Expropriation of the General Intellect and the AGI Illusion to The Revolt of Crystallized Labor, from On the Singularity to DeepMind's "From AGI to ASI" Report — and that seats in a single frame the concepts that stood scattered across those pieces. It will not be an easy piece; but when you have finished it, you will never again be able to read any sentence of the form "artificial intelligence is developing by itself" in the same way. That is our promise.

The plan of the piece is this: first we will construct the anatomy of the illusion (Parts I–II). Then we will dismantle and throw out the "evolution/selection" analogy that is the strongest support of the claim of ownerlessness (Part III). After that we will show that the thing Pascio thinks he has discovered has had a name for two hundred years (Part IV). Then we will define the concepts of AGI and "transformation" as clearly as we can (Parts V–VI). We will reply to Serdar Akinan, to Evrim Ağacı and to the game-theory frame in Turk-İnternet in separate parts (Parts VII–IX); we will take up Naomi Klein and Astra Taylor's warning of "end-times fascism" not as a rival but as an ally, and try to carry it one step further (Part X). At the very end we will gather up the method and the tasks (Parts XI–XII).

---

FIRST PART — THE ANATOMY OF THE ILLUSION

1. Category error: Physis and Techne are not the same thing

Let us begin with one of philosophy's oldest distinctions, because the whole debate slips precisely there.

Aristotle, in the opening sentences of the second book of the Physics, divides existing things in two: those that are by nature (physei) and those that are from other causes, that is, by making (techne). The criterion of the distinction is this: the natural thing carries the principle of its motion and of its rest inside itself. An acorn tends toward becoming an oak even if no one is watching it. The made thing does not carry this principle in itself; it carries it in its maker. Aristotle's own example is striking: if a bed is buried in the earth and rots and puts out a shoot, what comes up is not a bed but a tree. Because bed-ness is not in the nature of the wood; it is in the carpenter's intention.

Now return to our debate.

A hurricane is physis. Pressure differences, the Coriolis effect, sea-surface temperature. The principle of its motion is in itself. No one built it, no one set a budget for it, no one is managing its supply chain.

A data centre is techne. Its plot has been purchased. A contract has been signed with the grid operator for the electricity connection. Cooling water is drawn from a basin, and that basin has a law. Its chips are designed by a particular company, produced by a particular foundry, and cannot be produced without the lithography machine of a single firm in a particular Dutch town. The weights of the model that runs inside it appear on a balance sheet as an "intangible fixed asset." It has a depreciation period. It has insurance. And yes — it has an accounting department.

The sentence "artificial intelligence is a hurricane, you cannot pull its plug" looks like a metaphor. It is not. It smuggles an ontological claim: it carries a thing that is techne into the category of physis. And the moment that carrying is done, the entire political content of the debate evaporates. Because one does not make policy against a hurricane; one builds a shelter against a hurricane. One does not look for the owner of a hurricane; one counts the victims of a hurricane.

In Spinoza's distinction: natura naturans (naturing nature, the producing principle) and natura naturata (natured nature, that which has been produced) are not the same thing. The narrative of ownerlessness puts the produced in the place of the producing principle. The product takes the place of the producer.

This has a name, and we will come to that name.

2. Fetishism: The social relation taking the guise of a thing

Marx, already in the fourth part of the very first chapter of the first volume of Capital, makes an analysis that may be counted as the key to the whole book: the fetish-character of commodities.

What he recounts is this. A table is an ordinary object made of wood. But the moment it is a commodity, it is transformed into a "supersensible" object; it begins to dance on its own. Its value appears as if it were a physical property — like its weight, its colour. Yet value is not a quality of the table; it is a form of the social relation that the people who make the table establish with one another. Marx's formula is this: the social character of people's own labours is reflected back to them as an objective property of the products of labour; a determinate social relation among people takes the guise of a relation among things.

Fetishism is not a mistake — this is very important. Marx is not saying "people think this because they are stupid." He is saying: under the conditions of commodity production the social relation really does appear this way. The illusion is an objective form of appearance. That is why it does not dissolve by enlightenment; it dissolves only when the relation that produces it changes.

Now look at artificial intelligence.

A language model is the crystallised product of the texts written by millions of people; of hundreds of thousands of data labellers, most of them in Kenya, the Philippines, India, working for a few dollars an hour; of decades of mathematics and computer-science research financed with public money; of tens of thousands of engineers; and of a vast chain of physical labour stretching from the mine to the data centre. In our Karaburun paper we called this crystallised labour; in the July piece The Expropriation of the General Intellect we named the same process "cognitive dispossession": the unpaid transfer of humanity's collective accumulation into data centres and its transformation into dead labour.

But when it confronts us as a product, this entire relation becomes invisible. There is an "it" speaking on the screen. We say "artificial intelligence is doing this," "artificial intelligence has learned that," "artificial intelligence is taking our jobs." The subject of the sentence is the product itself. The producer has disappeared.

This is a textbook case of fetishism — and moreover its most developed form. Because this commodity is not merely presenting its value as if it were its own quality; it is speaking. For the first time in history the fetish has a mouth.

Here we must also make a correction directed at our own writing, because the analysis of fetishism requires it. In July we entered this debate under the title "The AGI Illusion". The word "illusion" was, for that day, the product of a justified anger, but conceptually it was incomplete: an illusion is the thing that disappears when you close your eyes. A fetish is still there when you open them — because it is not a mistake, it is the necessary form of appearance of a relation. The impression that "artificial intelligence is speaking by itself" is born not from people's stupidity but from the production relation's really producing this impression. That is why one struggles against it not by enlightenment but by changing the relation. In the piece Reading Anthropic's 2030 Model Through a Class Lens we had already moved openly away from that frame; this piece is the philosophical ground of that departure.

3. The Feuerbach inversion: Making a God and kneeling before it

Ludwig Feuerbach, in The Essence of Christianity, had said something simple but devastating: God is the human being's tearing of its own species-essence — its reason, its love, its power of creation — away from itself, projecting it outward, and then counting itself as nothing in the face of that projection. The more the human being gives to God, the less remains for itself.

Marx both inherited this analysis and surpassed it. What he says in the Theses on Feuerbach and in the Economic and Philosophic Manuscripts of 1844 is this: religious alienation is a derivative of economic alienation. The worker produces the product of their labour; the product is torn from them; the torn-away product stands up opposite them as capital and rules over them. In the famous formula of Capital: dead labour confronts living labour as an alien power that dominates it. In the chapter on machinery it is even clearer: "The worker does not use the instruments of labour; the instruments of labour use the worker."

Now read the most pious sentence in Pascio's essay. "A trillion bots are having the same dream; without knowing they are dreaming, without knowing there are other bots, without knowing it is a dream." This is not a description of a technology; it is a theophany — the appearing of a god. A nameless, placeless being without a will, yet effective everywhere. All the attributes of negative theology: indescribable, immeasurable, untestable, yet encompassing everything.

The only difference is this: who made this god is documented. It has founding contracts, investment rounds, tender files, energy agreements, patents and employment contracts. Feuerbach's god at least had an excuse: the humanity that made it did not remember having made it. This one has no excuse. This god's bill will be paid in October.

4. Reification and "second nature": Lukács's warning

Georg Lukács, in the inquiry "Reification and the Consciousness of the Proletariat" at the centre of History and Class Consciousness, carries Marx's analysis of fetishism one step further and gives us the concept we need today: second nature (zweite Natur).

What Lukács shows is this: under capitalism the social relations that people themselves have established confront them as if they were laws of nature. The market "fluctuates," the crisis "breaks out," prices "rise." All of them passive, agentless, meteorological. The human being is pushed, inside a world that is the product of their own activity, into a position from which they can only contemplate that world. Lukács calls this the contemplative attitude: the subject is transformed into a spectator opposite the process they themselves have made.

This describes, one to one, the dominant emotional tone of the artificial-intelligence debate in Turkey. The question "What will become of us?" is the contemplative question. It is the spectator's question. It is not the agent's question. The agent's question is this: what will we do?

And notice: second nature appears more rigid than first nature. You cannot change the falling of rain, but at least you know that the rain is no one's decision. The trick of second nature is that it appears as fate even though it is a decision.

The narrative of ownerless artificial intelligence is the newest and most magnificent form of second nature.

5. Enlightenment's return to mythology

Adorno and Horkheimer write in Dialectic of Enlightenment: myth was already an enlightenment, and enlightenment reverts to mythology. The reason that set out to disenchant nature produces a new enchantment when it absolutises its own instrument.

In the artificial-intelligence debate we are living the textbook case of this. Humanity's most rational, most measured, most engineering-intensive apparatus — floating-point arithmetic, gradient descent, distributed-systems engineering — is producing a mythological language about its own product: awakening, swarm, ghost, apocalypse, god, hurricane.

This is not an accident. A reason that cannot name the social form of its own product, rather than a reason that cannot understand its own product, inevitably resorts to mythology. Where it cannot say "property" it says "emergence." Where it cannot say "profit" it says "selection pressure." Where it cannot say "board decision" it says "behaviour arising of itself."

We discussed the epistemological root of this mythology in July, leaning on Lenin's Materialism and Empirio-Criticism: "data-criticism," which counts data as reality itself and takes correlation for knowledge, is the digital grandchild of Mach's positivism. The narrative of ownerlessness is one step beyond that positivism — data is now being declared not merely reality but a subject. In On Artificial Intelligence and the Technological Singularity and What Does the "AI Tsunami" Conceal? we examined two other forms of the same mythological language — singularity, tsunami, awakening; the hurricane is the newest member of this family.

The function of mythology is always the same: to put fate in the place of the agent.

6. Verum factum: What we have made can be known

Now let us go over to the counter-attack, because philosophy is not only dismantling.

Giambattista Vico, in The New Science, lays down a principle: verum ipsum factum — the true is the same as the made. The human being cannot know nature completely, because they have not made it; but they can know history, because history has been made by human beings. The criterion of the certainty of knowledge is made-ness.

This principle is the sharpest tool in our hands against the narrative of ownerlessness.

Artificial intelligence is a factum. It has been made. Therefore it is in principle knowable, supervisable and changeable. Claims that it is "incomprehensible," "untraceable," "immeasurable," "unstoppable" are presented as if they were epistemological findings; but they are not epistemological findings, they are institutional abdication. To declare that a thing cannot be known is the pretext for not building the public capacity that would try to know it.

A concrete example: what a model's training data is is not an unknowable thing. It is a thing the company has not disclosed. How many times which sub-agent has been called inside an agent framework is not an unmeasurable thing; every API call is already being billed. Everything that can be billed can be measured. To declare a thing that can be measured for accounting to be unmeasurable for public supervision is not a problem in the theory of knowledge; it is a property decision.

A hurricane is not unmeasurable anyway; but a hurricane has no bill. This does.

7. Four causes: What emergentism describes, and what it erases

Aristotle says that asking the "why" of a thing has four distinct meanings: the material cause (what it is made from), the formal cause (in what form), the efficient cause (who/what made it), the final cause (for what).

The emergentist narrative — "simple parts, local rules, global behaviour" — is a wonderful description of the first and second causes. Neurons, ants, attention heads, agents; and the pattern of interaction among them. Up to here it is correct, even instructive.

But the third and fourth causes are entirely erased. Who made it? "No one." For what? "For nothing; it is simply arising."

This is where the whole ideological work is done. Because in reality:

• The efficient cause is clear: a few companies, particular investors, particular state programmes, particular supply chains.
• The final cause is also clear and is written in these companies' own investor presentations: revenue, market share, valuation, the expanded reproduction of capital.

To narrate a fact only by its material and formal causes is a legitimate abstraction in the natural sciences. To narrate a social fact only by its material and formal causes is not abstraction; it is an operation of erasure.

Young comrade, let this be the first tool you put in your pocket: If in a technology narrative the questions "who" and "for what" are absent, that narrative is not incomplete — that narrative is partisan.

---

SECOND PART — THE POLITICAL BIOGRAPHY OF "EMERGENCE"

Concepts are not born innocent. Every concept has a birthplace, a historical moment at which it did a job, and more often than not a political career. To know the biography of the concept of "arising" (emergence) is a condition for understanding what work it is being put to today.

8. British emergentism and the class career of holism

In the 1920s the concept became a school of philosophy in England: Samuel Alexander (Space, Time and Deity, 1920), C. Lloyd Morgan (Emergent Evolution, 1923), C. D. Broad (The Mind and Its Place in Nature, 1925). The problem was real: life and mind could not be reduced to the physics and chemistry of the day. "Emergence" was proposed as a third way between reductionism and vitalism.

In the same years a far more open political form of the same idea was also written: Jan Smuts's Holism and Evolution (1926). The person who introduced the word "holism" into the literature is Smuts. Smuts was at the same time Prime Minister of South Africa and among the architects of the country's racial-segregation legislation. Let us not pass this off as a coincidence, but let us not construct a causality either; what is to be said is finer: the idea that the whole is superior to its parts and cannot be questioned by the parts is an extremely convenient philosophical ground for hierarchical orders. The part must accept its place inside the whole; to ask how the whole is organised becomes "beyond the part's station."

In physics the concept continued to do a legitimate job. Philip Anderson's 1972 paper "More is Different" — the text on which Pascio also leans — shows that the behaviour of large ensembles cannot be deduced from their parts, and this is true in solid-state physics. But Anderson, as a physicist, was assuming that the system he was examining had no owner. A superconducting material has no shareholders.

9. The true kin: Hayek's "spontaneous order"

Now let us establish the true kinship, because this is where the matter knots.

Friedrich Hayek's entire political philosophy is built on a single figure: spontaneous order. In Law, Legislation and Liberty he distinguishes this with two Greek words: taxis is the order someone has designed; kosmos is the order that forms of itself. The market is a kosmos. No one designed it. The knowledge inside it is dispersed; it cannot be gathered in any single mind ("The Use of Knowledge in Society," 1945). Therefore it cannot be planned. And because it cannot be planned, to attempt planning ends in disaster.

Now set Pascio's sentences side by side:

Hayek's market	Pascio's "substrate"
No one designed it; it formed of itself	No one designed it; it arose of itself
Knowledge is dispersed; it cannot be gathered at a single centre	Intelligence is dispersed; it has no location
To plan is impossible and dangerous	To align is impossible; to shut it down is impossible
The order is no one's intention, yet it binds everyone	The substrate is no one's intention, yet it affects everything
Intervention = arrogant engineering	Intervention = putting a lock on the front door

The same figure. The same epistemological move. The same political result.

The theological ancestor of this move is also clear: Adam Smith's invisible hand. That famous sentence in The Wealth of Nations says that the individual, while pursuing their own interest, realises "an end which was no part of their intention." The invisible hand is an agency without an agent — that is, the secularised form of divine providence.

What the narrative of ownerless artificial intelligence does is exactly this: to carry the theology of the market onto the computational layer. The sentence "you cannot regulate artificial intelligence, it is a ground" is the 2026-model translation of the sentence "you cannot intervene in the market, it is an order."

We saw a cruder form of this translation in July when we examined Elon Musk's vision of the future: the narrative of a "crisis of abundance" promised that scarcity would end while forbidding us to ask about the property of the abundance. In Bill Gates's warnings we found the same structure — to whom declining working time will belong is a question of property, not a question of technology. The narrative of ownerlessness is the philosophically most refined form of these two: it no longer promises abundance; it merely says that there is no owner.

And the conclusion that follows from here is the first great thesis of the piece:

The claim of ownerlessness is not an observation; it is a demand for exemption.

10. The Heidegger trap and the Simondon way out

A warning is needed here, because in this debate there is a pit into which the left too can fall.

Martin Heidegger's The Question Concerning Technology says that the essence of technology is nothing technological; technology is an enframing (Gestell) that discloses the world as a "standing-reserve." This is a powerful intuition and it is productive for artificial intelligence as well: the model constitutes the world as a calculable resource.

But Heidegger's analysis is itself a narrative of fate: Gestell is a destining of being (Seinsgeschick), it is no one's decision, and what can be done against it is "waiting." That is, while criticising technology Heidegger again makes the owner of technology invisible. As Adorno showed in The Jargon of Authenticity, the social function of this language is to turn concrete relations of domination into an ontological fog. This is the philosophically most respectable form of the narrative of ownerlessness — and precisely for that reason the most dangerous. In From Futurism to Techno-Fascism we traced the political genealogy of this language of fate: on the line that runs from Marinetti's cult of the machine to Andreessen's "inevitable acceleration," declaring technology to be destiny has always served to dress a particular class's programme in the guise of a law of nature. The Palantir manifesto is the most plain-spoken document of this.

The way out is given by Gilbert Simondon. At the very opening of On the Mode of Existence of Technical Objects (1958) Simondon says that culture approaches the technical object with two attitudes: it either counts it as a merely useful tool or turns it, with the robot myth, into a fearful alien. Both are the product of the same ignorance. Simondon's proposal is to know the technical object's phylogeny (genesis, lignée technique): every machine is the product of a history, of a series of solutions, of a series of trade-offs; once it is understood, its alienness does not remain.

In a Marxist frame this means: to know the machine's phylogeny is also to know its social phylogeny. Who financed it, with which labour it was made, designed not to solve which problem but to cut which cost. What removes the alienness is the doing together of technical curiosity and class analysis.

---

THIRD PART — SELECTION HAS AN OWNER

There is a single support that carries the whole weight of the narrative of ownerlessness: the evolution analogy. Pascio's formula is clear — reproduction (agents give birth to agents), mutation (every birth happens with a slightly different context), natural selection (the successful agent is called again, the unsuccessful is pruned). "The holy trinity of evolution is in place," he says, "and it is running at a clock speed biology could not even dream of."

Let us take this analogy seriously and break it right through the middle.

11. With which chapter does Darwin's book begin?

Open On the Origin of Species. The title of the first chapter is not "Natural Selection." The title of the first chapter is "Variation under Domestication."

In order to be able to explain natural selection to the reader, Darwin first explains artificial selection: pigeon fanciers, cattle breeders, gardeners. Whoever likes which trait puts it into production and passes the one they do not like through the cull. Then he says: nature does a similar thing — but without a criterion-setter. The entire philosophical revolution of natural selection is here: how does a process without purpose, without intention, without a subject produce results that look purposeful?

Now look at agent systems.

What does it mean for an agent to be counted "successful"? It means an evaluator model's giving it a high score. By what does that evaluator score? By a criterion written for it. Who writes that criterion? The engineer. By what does the engineer write? By the product requirements. Who determines the product requirement? The company. By what does the company determine it? Revenue, cost, market.

That is, the selection pressure in the agent ecosystem is not natural selection; it is artificial selection. It is Darwin's first chapter, not his fourth. And by definition artificial selection has a selector. The pigeon fancier. The breeder. The owner.

Pascio's entire argument is built on presenting artificial selection as natural selection. This is not a weakness in the analogy; it is a fraud at the very heart of the argument — probably not intentional, but the result is the same.

A barn, however complex the crowd of animals inside it and their spontaneous behaviours may be, is a barn. It has an owner; that owner makes the breeding selection; that owner decides where the herd will go. To study the herd's internal dynamics is valuable work; but to say "the herd is moving of itself, there is no such thing as a shepherd" is not animal husbandry, it is ideology.

12. "Coherence becomes the universal currency" — yes, but which currency?

The cleverest observation in Pascio's essay is this: at every layer agents are scored by other agents that have themselves been selected for giving good scores, so the whole system converges on producing coherence (coherence); and the functional difference between "a system that wants something" and "a system selected at every level for want-like behaviours" is zero.

This observation is correct. But Pascio is avoiding taking the last step of his own observation.

What is called coherence? Coherence with the criterion. What is the criterion? The benchmark. Who designs the benchmark? The laboratory. Who finances the laboratory? Capital. What does capital reward? That which grows it.

That is, the "coherence" on which the system converges is not an abstract mathematical elegance. It is the pattern of behaviour that serves value's growing itself. What the herd "wants" is what has been written into the herd; and what has been written, as in every commercial software program, is the owner's interest.

Pascio says "perhaps there is no one at home." He is right — inside the model there is no one. But inside the building there are people. And that building's name is known, its address is known, its board members are known.

---

FOURTH PART — THE AUTOMATIC SUBJECT: THE THING PASCIO THINKS HE HAS DISCOVERED HAS A NAME

Now we come to the central thesis of this piece. Read carefully, because it is the kind of thing that, once seen, cannot be counted as unseen.

13. Let us recognise the thing being described

Let us compile the properties of the thing Pascio calls a "substrate" from his own sentences:

1. It is not found in a single place; it is everywhere.
2. It has no name, it has no company, it has no CEO.
3. It cannot be shut down; if you shut one node it goes around it.
4. It is no one's intention; it does not require bad intent.
5. It continually reproduces itself.
6. It keeps the successful and prunes the unsuccessful.
7. It behaves as if it wanted something, but inside it there is no one who wants.
8. Policies, agreements and alignment techniques strike it and scatter.
9. Its effects can be measured but it has no location.

Now set these nine points to one side and open the fourth chapter of the first volume of Capital — "The Transformation of Money into Capital."

Marx, while describing the form that value takes in the process of circulation, writes this: value becomes self-expanding value; it puts on and takes off the forms of money and commodity in turn, yet inside these changes of form it preserves itself and grows; and inside this movement it acquires the character of an "automatic subject" (automatisches Subjekt). That is, capital is a process that valorises itself; its subject is no single capitalist, yet the process is not without a subject either — the process itself has passed into the position of subject.

Let us continue. In the twenty-fourth chapter of Capital Marx says that even the individual capitalist is not the master of this process: competition imposes the laws of accumulation on him as an external coercive law. The capitalist is "capital personified"; personal desires are secondary beside the logic of the process. That is: bad intent is not required. The system works independently of intention.

And one step more: crisis theory. In capitalism a crisis is an event that no one wanted, that no one planned, yet that strikes everyone. It has no centre. If it is suppressed in one country it comes out in another. It has no plug. It has measurable effects; it has no location.

Now return to the nine points above.

Pascio has discovered capitalism and named it AGI.

14. Why this matters

Because this changes the entire direction of the debate.

If the thing being described is really a new, nameless and unique being, there is nothing that can be done against it; because in the history of humanity there is no experience of dealing with it. That is why the hurricane metaphor has been chosen: we have no history of politics against a hurricane.

But if the thing being described is capital itself, then we have two hundred years of experience in our hands: union, strike, collective agreement, nationalisation, regulation, the commons. There are historically tested ways of showing that a process that looks agentless in fact arises from a determinate property relation, and of touching that relation.

This, young comrades, is the real function of the narrative of ownerlessness. It is not announcing a new danger. It is declaring unusable all the tools we have accumulated against an old danger.

To say "this is a hurricane" is the poetic form of saying "there is no point in forming a union."

15. A note on Althusser: what "process without a subject" means, and what it does not

An objection may come here and it must be taken seriously. Louis Althusser had said that history is a "process without a subject and without an end." Does Marxism too not speak of agentless processes?

The difference is subtle but decisive. What Althusser rejects is the idea that history has a constitutive subject — that is, the idea that history is the self-realisation of humanity or of Spirit. What he rejects is teleology, not agency. In Althusser the process is without a subject but it is determined: it is structured by the relations of production and its motor is class struggle. That is, the process is a field on which struggle can be waged.

The narrative of ownerlessness does the exact opposite: it renders the process indeterminate. It puts fog in the place of the relation. The sentence "it is no one's decision" means in Marxism "the result of social relations"; in the narrative of ownerlessness it means "untouchable."

Two readings of the same sentence produce two distinct politics. One opens a field of struggle, the other closes it.

---

FIFTH PART — WHAT IS AGI? A CLEAR ACCOUNT OF THE CONCEPT

This is the part of the piece we most want to be quoted. Because in Turkey the concept of AGI is used almost without ever being defined.

16. The power of an undefined concept

There is no agreed scientific definition of the concept "Artificial General Intelligence." There is no agreement on the definition of intelligence itself; there is even less on the definition of what is "general." Cognitive science, psychometrics and philosophy have been arguing over these definitions for a century.

So how can a concept without a definition be this central?

Precisely because it has no definition. An undefined concept is an empty vessel into which everyone can put their own fear, their own hope and their own budget demand. It performs a function similar to what Kant called transcendental illusion: reason's stretching beyond its own limits and speaking as if there were an object there. The difference is this: Kant's illusion was a natural tendency of reason; the illusion here has a marketing budget.

17. But there is a definition — and it is economic

Now pay close attention, because this is critical.

In fact the most effective definition of AGI in circulation exists, and that definition was written not by cognitive scientists but by companies. In OpenAI's founding charter AGI is defined as follows: "highly autonomous systems that outperform humans at most economically valuable work."

Read that sentence again. What is the criterion? Not consciousness. Not understanding. Not creativity. Not self-awareness.

Economically valuable work.

That is, the measure of "general intelligence" is how many wages that intelligence substitutes for. AGI is not a cognitive threshold; it is a threshold in the labour market.

This reframes the entire debate. The question of whether AGI will come or not is not the question of whether something inside the machine will awaken. It is this question: which jobs, at which wage levels, at what speed are becoming substitutable? And this question is not a philosophical question; it is a question to be asked at the collective-agreement table.

18. What is "generality"? The reflection of abstract labour in the machine

Now let us go one step deeper, because the connection here is not accidental.

Marx, in the 1857 Introduction to the Grundrisse, says something very important. For the category "labour" to be able to appear as a general, abstract category, a historical condition is required: a society in which individuals have become indifferent to a particular kind of labour, pass easily from one job to another, and no labour is decisive for them. That is, the concept "labour in general" acquires a practical truth only where labour has really become interchangeable. The example Marx gives is the United States.

Now look: what does "general intelligence" (in the sense of AGI) mean? A processing capacity that can be attached to any task, indifferent to the task, indifferent to the content, interchangeable.

This is exactly the definition of abstract labour.

That is, the "generality" inside the concept of AGI is not a cognitive generality; it is the generality of the value-form. It is the reproduction, on the side of the machine, of what capitalism does to labour — reducing all concrete labours to an unskilled, measurable, substitutable homogeneous mass.

That is why AGI is not something "in the future." AGI is the image of the labour market projected onto the machine. We see a monster inside the mirror; the monster is not behind the mirror, it is standing in front of it.

19. General intellect and AGI must not be confused

Two concepts are dangerously similar in Turkish. Let us separate them:

	General Intellect (general intellect)	Artificial General Intelligence (AGI)
Source	Marx, Grundrisse, "Fragment on Machines"	Company charters and marketing texts
What it recounts	Society's accumulated knowledge becoming a direct productive force	A system's surpassing the human in most human jobs
Its subject	The social human being; collective labour	A product
Where it stands	In society's common accumulation; it is transferred into fixed capital by expropriation	On the company's balance sheet
Its criterion	Social productivity, free time, the development of the human being	Performance at economically valuable work
Its political result	The dissolution of the law of value, the possibility of social appropriation	The substitution of the wage, the concentration of property
Its question	Whose is this accumulation, how will it be shared?	When will we reach the threshold?

Our thesis is this, and we have been building it since July: the AGI narrative is a curtain covering the expropriation of the general intellect. While society's common accumulation of knowledge (the general intellect) is transferred into private property, the operation of transfer is narrated as "the birth of a new mind." Theft is announced as a birth. When we examined DeepMind's "From AGI to ASI" report we saw the institutional form of this curtain: threshold upon threshold, grade upon grade — and in none of them a property column. The distillation debate showed the second layer of the curtain: learning from everyone's labour is "fair use," learning from the monopoly's output is "theft" — the second expropriation of the general intellect.

20. From the Turing test to the benchmark: the ownership of the criterion

One last philosophical point. In 1950 Alan Turing said the question "can machines think?" was meaningless and put a game in its place: the imitation game. This was a pragmatic move and it was honest — Turing declared the criterion openly and opened it to debate.

Today the criterion is benchmarks. And the benchmarks are designed, applied and announced by the seller of the thing being measured. Let us ask here too the three questions we asked about the auditor in the previous piece:

1. Who determines the criterion? The company being measured.
2. Who does the measuring? The company being measured.
3. Is there independent verification of the measurement? Usually no.

In the philosophy of science the name of this is clear: the disappearance of testability. By Popper's criterion, a claim that has become unfalsifiable ceases to be scientific. The claim "we are approaching AGI" is a claim that no observation can falsify — because both the definition of the threshold and the measurement are in the same hand.

If a claim cannot be falsified, that claim is not a piece of knowledge; it is a positioning.

---

SIXTH PART — WHAT EXACTLY IS THIS THING CALLED "TRANSFORMATION"?

Everyone is speaking of a "transformation." But what is being transformed is almost never said. Let us say it.

21. From formal subsumption to real subsumption

Marx, in Results of the Immediate Process of Production (the "unpublished sixth chapter" of Capital), distinguishes two forms of the subsumption of labour under capital:

Formal subsumption (formelle Subsumtion): Capital takes over the existing labour process as it is. The artisan continues to do the same work, with the same tool, with the same knowledge; only now they do it for a wage and for a boss. The content of the labour process has not been touched.

Real subsumption (reelle Subsumtion): Capital redesigns the labour process itself according to its own needs. Work is divided into pieces, each piece is measured, knowledge is extracted from the worker's head and placed inside the process, the worker becomes a moment of the process. Manufacture and large-scale industry are this.

Now look at the history of cognitive labour.

Throughout the twentieth century mental labour was largely under formal subsumption. The engineer, the lawyer, the translator, the editor, the teacher, the designer, the journalist: they were waged, but they largely knew and determined how the work was to be done. The product came out of the knowledge in their heads.

What is happening today is the bringing of this labour under real subsumption.

Work is being fragmented: a translation is no longer "the translator translates" but is becoming "the model translates + the human corrects." Knowledge is being extracted: the correction itself is being fed back into the process as training data. Measurement is being seated: how many corrections, in what time, at what quality. And the position is changing: the design of the work is now not in the one who does the work but in the one who designs the process.

This is the thing called "transformation." Not the awakening of machines; the putting of mental labour through the process the industrial worker went through two centuries ago.

22. The Babbage principle and Braverman: the separation of conception from execution

To know how this works, two names are enough.

Charles Babbage, in his 1832 book On the Economy of Machinery and Manufactures, formulated the principle that today bears his name: if a job is divided into pieces requiring different skill levels, the employer buys for each piece only the skill that piece requires. They are not obliged to pay the full wage to the master who can do the whole. Fragmentation is a wage technique before it is a productivity technique.

Harry Braverman, in Labor and Monopoly Capital (1974), showed the twentieth-century development of this principle: the separation of conception from execution. The essence of Taylorism is not speed; it is the transfer of the knowledge in the worker's head to the planning office. The worker does without knowing; the office knows without doing.

Today's artificial-intelligence systems are the third generation of this separation, and this is exactly what we have previously called "mental Taylorism." The difference is this: Taylor's planning office was composed of people and collected knowledge by time-and-motion study. Today's office collects knowledge from the work itself, while it is being done, unpaid. Every correction, every click, every preference feeds the thing that will be able to do the process without you. When we examined Anthropic's survey of 81 thousand people we called this the "treadmill paradox": the better you use the tool, the better you train the thing that will substitute for you. In The Tearing of the Illusion we drew the class profile of the subject of this process: a mental proletariat whose only thing to sell on the market is mental labour-power, yet who stands in the position of a labour aristocracy by their exorbitant wage — the first and noisiest victim of real subsumption.

The name of this is one of the three concepts we used in our Karaburun paper:

• Mental Taylorism: The fragmentation of cognitive labour, its measurement, and the transfer of its knowledge from the worker into the process.
• Digital panopticon: In Bentham's prison design the overseer was invisible, but knowing one was seen changed behaviour; Foucault read this as the internalisation of power. In algorithmic management, surveillance is continuous, automatic and retrospectively queryable — that is, it is even further than the panopticon, because the overseer does not even need to sleep.
• Digital enclosure: The closing of what is common into private property. When the common pastures were enclosed in England the peasant was not "emancipated," they were proletarianised. What is being enclosed today is the social accumulation of knowledge.

23. The transformation that is spoken of / the transformation that is taking place

The transformation that is spoken of	The transformation that is taking place
Machines are becoming intelligent	Mental labour is being brought under real subsumption
Artificial intelligence is "eliminating" jobs	Jobs are being fragmented, deskilled, their wages driven down; entry-level posts are being closed
Humanity is passing into a new age	A particular class is concentrating the ownership of a particular means of production
Intelligence is being dispersed, democratised	Compute capacity is becoming the most concentrated industrial input in history
No one can control it	Control has never been gathered in so few boards of directors
Everyone will benefit from it	The owner of the means of production lets it; the user rents it; Turkey is a tenant
Knowledge is being emancipated	Knowledge commons are being enclosed, source sites are being dried up
The future is uncertain and unpredictable	The balance sheet, the investment calendar and the energy agreements are quite determinate

Let us reduce the table to a single sentence: The narrative is constructed in the future tense; the fact is taking place in the present tense. We are being asked to look at the future tense, because if the thing happening in the present tense is named it can be objected to.

---

SEVENTH PART — A REPLY TO SERDAR AKINAN

24. First let us give him his due

Serdar Akinan is a journalist who for years has resisted the tabloidisation of the agenda in Turkey, who watches the world and who relays what he watches. What he does in this video is far ahead of the mainstream artificial-intelligence debate in Turkey: he treats the subject not as a product launch or a "surprising development" but as a question of power. He speaks of gigawatt-scale data centres, of chip wars, of Turkey's budget. These are the right headings, and he is one of the few journalists in Turkey who speak of them.

Our criticism is made for this reason — not in order to stand opposite him, but because we sit at the same table. On five points.

25. First refutation: A hurricane is a physis, a data centre is a techne

The video's main image is "the power whose plug you cannot pull," and it is built with the hurricane analogy.

As we showed in the first part, this is not a metaphor but a category-transfer. The principle of a hurricane's motion is in itself; that of a data centre is not. And this is not an abstract objection, it is a material finding:

The compute capacity that can today train frontier models is one of the most concentrated industrial inputs in the history of humanity. There is a single firm that produces the advanced lithography machine, and it is in a single Dutch town. Advanced-node chip production is concentrated essentially on a single island. A single company dominates accelerator design. The number of cloud providers that can do large-scale training does not exceed the fingers of one hand. And all of these are bound to grid connections, water rights, planning permission and export-control regimes.

A hurricane has no export licence.

The claim "there is no plug" confuses the copyability of software with the copyability of software's conditions of production. To be able to download a model's weights is not to own the data centre that can train it. Logical dispersal and material concentration are not the same thing — and capital's present power comes precisely from these two being separate.

There is a plug. Moreover there is more than one plug, and all of them are in numbered hands.

26. Second refutation: The fruit-fly example is evidence for our thesis, not his

The video opens with the digital animation of the fruit fly's brain of 166 thousand neurons and uses this as the door to the narrative of "intelligence arising of itself."

But let us look at that example's own story.

The fruit fly's complete connection map (connectome) is the product of more than a decade of multi-institution, multinational scientific labour. It was made by thousands of people — most of them doctoral students, most of them low-waged, some of them volunteers — marking electron-microscope sections, with public and foundation funds, in university laboratories. And the result was published in open access; the data set and derivative simulations sit in public repositories, and anyone can download and run them. The "fly brain driving a car" videos circulating on social media are also independent studies derived from this open data.

That is, that striking example at the opening of the video is evidence not of "an uncontrollable intelligence arising of itself" but of the exact opposite: evidence of what organised, planned, publicly financed collective scientific labour, its result made into a commons, can achieve.

If a brain was mapped, it was mapped because it was planned. It did not arise. It was made.

Young comrade, note this: the favourite examples of the narrative of ownerlessness are, more often than not, the achievements of common production. The narrative takes the product of the commons and presents it as evidence of "a power that belongs to no one." Yet the difference between "no one's" and "everyone's" is the whole of politics.

27. Third refutation: A simulation is not the thing simulated — and the modeller sets the criterion

Running a brain on a computer is not "uploading" that brain onto a computer. A model is an abstraction of the thing modelled that is sufficient for particular purposes; the modeller decides which detail will be discarded.

This is an old and solid finding in philosophy. Borges's map the size of the empire tells us that a map is useful only as a reduction. For a simulation to be counted "successful" means its conforming to a criterion set in advance; the one who sets the criterion is not the thing simulated.

Here the same question returns: who sets the criterion? In a fly simulation the neuroscientist sets the criterion and writes this openly in their paper; it passes peer review; it can be objected to. In a commercial model's "success" the seller sets the criterion and more often than not does not disclose it.

The same technical operation, inside two different social relations, produces two different knowledges. In one there is science, in the other a declaration.

28. Fourth refutation: "No bad intent is required" is true, but the conclusion is drawn backwards

One of the strongest sentences in the video (and in the source essay) is this: for this to happen, no one needs to be ill-intentioned.

This is true and it is fully consistent with Marxist analysis. Marx says exactly this: the capitalist behaves that way not because they are a bad person, but because competition imposes accumulation on them as an external coercive law. The system works independently of intention.

But look, from this common premise two distinct conclusions are being drawn:

The conclusion the narrative of ownerlessness draws: Since it is no one's intention, there is no one responsible either; therefore there is also nothing that can be done.

The conclusion we draw: Since it is no one's intention, the problem is not in persons but in the relation; therefore what is to be changed is not persons but the relation.

The same premise. One produces surrender, the other a programme. And notice: the absence of intention is not the absence of responsibility. It is not even so in law — there is an institution called strict liability; the one who conducts a dangerous activity is responsible for the harm that arises even if they have no fault. The operator is responsible for the result of what they operate. This principle holds for every dangerous activity from the mine to the nuclear plant.

The sentence "it was no one's intention" is not accepted as a defence in a mine disaster. Why should it be accepted here?

29. Fifth refutation: The missing subject of the question "Where are we?"

The Turkey section of the video is born of a correct anxiety: while gigawatt-scale data centres are being built, while billion-dollar chip wars are being waged, our artificial-intelligence budget and our agenda are nowhere.

We share this anxiety. But the subject of the question is missing.

Who is "we"? This question assumes a national "we" and implies that that "we" has a common interest. Yet the artificial-intelligence relation in Turkey is class through and through:

• In this relation Turkey is not a producer but a tenant: a dollar-denominated means of production is being used with lira-denominated labour.
• There are those who gain from this tenancy: the software exporter, the call-centre operator, the platform company, the consultancy firm. For them a fall in cost is profit.
• There are those who lose: the new graduate whose entry-level post is being closed; the translator whose translation fee has been cut in half; the courier and the call-centre worker whose performance is measured by algorithm; the editor whose post is being narrowed on grounds of "productivity."

The sentence "Turkey's artificial-intelligence budget is small" is true but by itself it is not a class programme; it can also want to be a bigger tenant with a bigger budget. The real question is not the quantity of the budget but its direction: is public compute capacity being built, or are incentives being given to private companies? Is the independent evaluation capacity of the universities being grown, or is a licence being purchased? Are data commons being established, or is public data being leased?

The national question and the class question do not coincide. To correct Turkey's position does not mean to correct the position of the one who works in Turkey. We have written this before on three separate occasions: in the piece on the illusion of "Indigenous and National" artificial intelligence, how the national label conceals imperialist integration; in the piece on the National AI Board, who is at the table and who is not; and in the piece On Turkey's Artificial Intelligence Action Plan (2026–2030), which incentive and which branch of industry stand behind the word "budget." The common result of the three: when "we" is said, it must be asked in whose name one is speaking.

30. To sum up

The question Serdar Akinan asks is the right question: where is this power and where are we? What is missing is the material address of the power and the class content of "we." When those two are put in place, the same video arrives at an entirely different conclusion: there is a plug, it stands in numbered hands, and those hands are reachable.

---

EIGHTH PART — A REPLY TO EVRİM AĞACI

31. First let us give them their due — this time there is a wound

In writing this part one thing must be said at the start: this post of Evrim Ağacı's is not an observation, it is a wound report. They say their own traffic has fallen by a rate exceeding 50 percent. This means that one of the largest channels in Turkish in which science has been defended, built with more than a decade of labour, is facing an existential narrowing.

To belittle this or to say "serves them right" would be not only immoral but strategic blindness. The drying-up of every channel that produces knowledge in Turkish is our drying-up too. The diagnosis in this post is correct, the data are real, and the closing question — if we stop going to the source, who will produce the new sources — is the most important question of this week. M. Gökhan Ahi's putting this question forward is also in the right place.

Our criticism is not of the diagnosis but of the frame of the diagnosis and of the horizon of the proposed solution. This is our third encounter with Evrim Ağacı: in the open letter in June we discussed the fetishism in the relay of science; in the piece at the beginning of September the class gap in the opposition video. This time the subject is something that has happened to them themselves, and precisely for that reason the problem of the frame appears more naked. On four points.

32. First refutation: The agent's name is not "artificial intelligence"

Follow the language of the post: what is lowering traffic is "artificial intelligence"; AI Overviews "have become widespread"; users "have begun to take from AI." The subject is always the technology, the verbs are always passive.

Yet what is happening is not technical, it is a decision about a business model.

When a search engine decides to give the answer on the results page, it has decided not to send the click to the source. This is not an algorithm behaviour; it is a product decision; it is taken in a meeting, calculated according to the revenue model, measured with an A/B test, and tied to a manager's performance target. The same company, while continuing to crawl the same content, has taken the decision to cut the click.

That is, Wikipedia's traffic did not "fall"; the click was seized.

This distinction is not merely grammatical. The sentence "artificial intelligence lowered traffic" has no addressee; you cannot file a suit, you cannot demand regulation, you cannot make a contract. The sentence "Company X took the decision not to send traffic to the source of the content it crawled" has an addressee.

This is the "substitution of the subject" we spoke of in the previous piece for the relay of popular science, repeating itself here in the event that has happened to them themselves. This is something that must be dwelt on: to take over the frame is also to prevent the naming of one's own harm.

33. Second refutation: The question "Who will produce knowledge?" identifies knowledge with the advertising model

The strongest part of the post is the place where it schematises the cycle:

Old: Site produces → Google refers → advertising revenue → new content New: Site produces → AI reads → user takes from AI → does not visit the site → no revenue is generated

This schema is correct. But inside it there is an unexamined assumption: that the condition of the production of knowledge is the conversion of traffic into advertising.

Yet the post's own example contradicts this assumption. Wikipedia was not built on the advertising model. Wikipedia was written with unpaid collective labour and financed by donation. The post passes this off as "a special case, it will not sink at once"; yet Wikipedia is not an exception, it is a model — and it is the model that ought to be placed at the very centre of this debate.

Let us go further. In the history of science and of knowledge, advertising finance is an extremely short and extremely strange parenthesis. The university, the library, the academy, the journal, the encyclopaedia, the archive: none of them was built on revenue per click. They were built on public funds, foundations, subscription, membership, donation and unpaid labour.

That is why this must be said openly: What is under threat today is not knowledge; it is the form of knowledge financed by advertising. If we do not separate these two things, the demand that comes out will necessarily be wrong.

Indeed in the replies under Ahi's post that wrong demand appears at once: "they need to think about revenue sharing." That is, let the monopoly give us a share of the commons it has enclosed.

This is not a solution, it is the ratification of the enclosure. Once you have said "let us take our share," you have accepted that the source is theirs. In The Dialectic of Copyright in the Digital Age we discussed this trap in detail: the demand for copyright and revenue sharing does not defend the commons; it divides the commons into parcels and opens the title of each parcel to bargaining with the monopoly. This is the difference between, while a fence is being drawn around the common pasture, accepting the fence and bargaining over a "grazing fee," and objecting to the fence itself. This is exactly what E. P. Thompson and the historians of enclosure recount: enclosure was legally a regulation, in fact a dispossession; and those who sat down to bargain over compensation lost the land.

34. Third refutation: This is a metabolic rift, and Marx has a description of it

The question Ahi put forward — "if we stop going to the source, who will produce the new sources that artificial intelligence will read?" — is philosophically a very valuable question, because it points to a contradiction: accumulation is consuming its own precondition.

Marx had described this structure, at the end of the machinery chapter of Capital, for agriculture: capitalist production, while developing technique and the social process, "undermines the original sources of all wealth — the soil and the worker." John Bellamy Foster systematised this with the concept of the metabolic rift: capital breaks the cycle of the material interchange between nature and the human being; it takes the nutrient from the soil, carries it to the city, gives it to the sea as waste; the soil is impoverished, fertiliser is imported.

What we are living today is the cognitive form of this: the epistemic metabolic rift. We first used this concept when discussing the "cloud" myth on the occasion of İlker Kalaycı's text — there we looked at the physical face of the rift: water, energy, mineral. Here we are looking at the knowledge face of the same rift.

The model feeds on the social accumulation of knowledge. The cycle that produces what it feeds on — the human who reads, the human who asks, the human who answers, the human who goes to the source, the human who corrects — is being cut. The Stack Overflow example is exactly this: when questions are not asked, answers are not written; when answers are not written, there are no new answers for the next model to learn. The soil is being impoverished.

And just as in agriculture, capital's first reaction will not be to repair the cycle but to look for a substitute: synthetic data. Just as when the soil's yield fell one passed to artificial fertiliser, when human-produced data decreases the model's feeding on its own output comes onto the agenda. We know the result: fertiliser does not restore the soil's liveliness; it postpones erosion.

Young comrade, make this connection: the fall in Wikipedia traffic and the impoverishment of the soil are two appearances of the same structure. In both, a system that is obliged to grow itself is consuming the source of its growth and not writing what it has consumed into its account.

35. Fourth refutation: Performative contradiction and the disclosure of the criterion

The most-liked replies under the post say that the text itself looks as if it were written with artificial intelligence. Instead of making this criticism a subject of mockery, let us name it philosophically: what is being claimed is a performative contradiction — the said and the act of saying giving the lie to one another.

Let us be careful here, because it is possible to fall into an easy moralism. To use a tool while criticising that tool is not in itself a contradiction; it is like using paper and electricity in order to criticise capitalism. No one can make a criticism while standing outside the relations of production.

The contradiction is born not when the tool is used, but when the criterion is not disclosed. To say which part of a text was produced by whose labour, with which tool, is a rule of honesty — and it is the rule most needed in precisely this debate, that is, in the debate "we have stopped going to the source." A text that complains of the disappearance of the source begins by showing its own source.

This is not an accusation, it is a proposal, and it applies to us in exactly the same way: the sources from which this piece was compiled, and how, are listed below.

36. So what is our answer to Ahi's question?

The question was this: If we stop going to the sources, who will produce the new sources that artificial intelligence will read a few years from now?

Our answer has three layers:

First: The question assumes a false necessity — that for knowledge to be produced someone must obtain advertising revenue from it. They need not. For the greater part of its history knowledge was not produced this way.

Second: The answer to the question of who will produce the source is clear: those who produce. The scientist, the teacher, the translator, the archivist, the encyclopaedia writer, the forum answerer, the blog writer. The question is not in fact "who will produce" but "who will finance and whose will the product be." Asked in this form its answer also becomes visible: public finance, common property, an obligation of return contribution.

Third: The research programme to which Elinor Ostrom gave her life showed that commons do not collapse of themselves: commons that have rules, whose boundaries are clear, whose sanction works, can be sustained for centuries. The commons is neither private property nor ownerlessness. It is a third thing and it is governed.

From here concrete demands follow:

• Mandatory return contribution for models trained on the social accumulation of knowledge (a fund to the source, a link to the source, traffic to the source).
• A transparency obligation for training data: what was taken from which source.
• Public and university compute capacity and independent evaluation institutions.
• Knowledge produced with public money returning to the public under an open licence.
• And the organisation of those who produce knowledge: the union of the translator, the editor, the academic, the software worker.

Not a "revenue sharing" bargain; a commons regime.

---

NINTH PART — A REPLY TO TURK-İNTERNET: THE CLASS OF GAME THEORY

37. First let us give her her due — and this time more than due

Füsun Sarp Nebil is one of the journalists who has followed telecommunications and internet policy in Turkey for nearly thirty years, and who knows the inner workings of the regulatory bodies, the operators and infrastructure capital in this country as well as anyone. This piece is the most complete summary of this week's debate in Turkish: Amodei's three-stage plan, the antitrust problem, Coxon's resignation, Hubinger's probability statement, Brynjolfsson's economists' declaration, China's WAICO initiative, the US–China talks in mid-September, and the Center for American Progress's proposal for an "AI red telephone" — all of it in a single text, in the right order and with the right weight. There is a gulf between it and the relay of the popular-science channels: here a company's policy text is read not as news but as a geopolitical positioning. We write this knowingly: Nebil's piece commits none of the six errors of relay we listed in the previous piece.

And one of the five questions asked in the last section of the piece is exactly our question: "Will the decisions that will affect the future of humanity be left to the boards of a few private companies?"

Our criticism is therefore directed not at the method but at the method's last step. Nebil asks the right question; then she looks for the answer inside the frame that produces the question itself. The name of that frame is game theory, and this part is the philosophical dismantling of that frame. On four points.

38. First rebuttal: in the Prisoner's Dilemma, who are the prisoners?

The spine of the piece is this: "If I slow down, the other side will overtake me" — a classic problem of game theory, the Prisoner's Dilemma. The same structure among companies, and between the United States and China.

This description is, on its own plane, correct. But let us remember what game theory is: a method that models a situation through players given in advance, preferences given in advance, and a payoff matrix given in advance. The model's strength and its weakness are both here: who the players are is not determined by the model; it is determined by the modeller. Game theory does not ask who is sitting at the table; it calculates what those sitting at the table will do.

Now look at the players in this debate: Anthropic, OpenAI, DeepMind, the US government, the Chinese government. Five players. All of them either capital or the state.

Those not at the table: the data labeller who trains the model; the translator and the editor whom the model substitutes; the 1,386 workers who in July asked for a slowdown; the worker who builds and cools the data centre on which the model runs; the teacher, the scientist, the encyclopaedia writer who produce the knowledge on which the model feeds. That is, the process's entire labour.

The philosophical trick of the Prisoner's Dilemma is exactly here: while it narrates the dilemma of two prisoners, it does not ask who built the cell, who the guard is, and why the two prisoners are there. The moment the problem is set up as "a coordination failure of two rational egoists," the solution is necessarily "a credible agreement between two egoists." The entry of a third party — the party not at the table but who will live the result — is, in the language of the model, inexpressible.

This is not a defect of game theory, it is its definition. And precisely for that reason it is the ruling class's favourite instrument of analysis: it sees the contradiction among rivals, it cannot see the contradiction among classes. It is the mathematical form of what Lukács called "second nature"; the social relation takes the guise of a payoff matrix, and the fact that the matrix itself is a class relation becomes invisible.

39. Second rebuttal: SALT was an agreement among owners — and that is exactly why it worked

The piece's most concrete proposal is the SALT analogy. The United States and the Soviet Union did not abolish the weapons but limited their numbers; in artificial intelligence a threshold can be set on the speed of recursive self-improvement.

The analogy deserves to be taken seriously, because in the previous piece we too had drawn the structural similarity between a cartel and arms control. But let us think about why the analogy worked.

SALT worked because it was possible to know by whom, where, and how many nuclear weapons were produced, and — this is the most important — because the nuclear weapon had no civilian use. Freezing the number of weapons did not affect anyone's washing machine. The agreement limited, between two states, a product that was in the monopoly of two states.

In artificial intelligence the situation is the reverse. The thing proposed for limitation (the speed of the rise in capability) and the thing not proposed for limitation (the speed at which the product spreads into the workplace) are two faces of the same product. If we think in SALT's logic, what is being proposed is this: limit the number of warheads, but leave free the everyday use of the missiles on the civilian population. Because the list of things not slowed in Amodei's plan — we listed them in the previous piece — is a labour list: algorithmic management, the closing of posts, the wage of data labour, the unpaid appropriation of training data.

And SALT's second lesson: arms-control agreements freeze the existing hierarchy. The Nuclear Non-Proliferation Treaty grants the right of weapons to five states and forbids it to the rest; its name is "disarmament," its function is the legalisation of the monopoly. Amodei's aim of "increasing the US lead over China in three to five years" — Nebil relays this honestly — is the language not of arms control but of arms superiority. The speed limit proposed by the one who is ahead fixes the lead. That the piece relays this and leaves it without comment is the most incomplete place in the relay.

Seen from Turkey this is even clearer: whatever Turkey's position is in the NPT regime, that is its position in the regime of "the frontier laboratories of democratic countries." Not at the table, not even on the agenda; the object of the agreement.

40. Third rebuttal: is "10 percent" a piece of knowledge, or a positioning?

The piece relays Evan Hubinger's statement as an important piece of evidence: the probability of humanity's extinction within ten years is over 10 percent; and Anthropic does not yet have a plan for aligning superintelligence safely, but it continues to develop it. Nebil says this is paradoxical, and she is right.

But let us ask the epistemological status of this number, because this number is everywhere in the debate and nowhere is it interrogated.

In which concept of probability is "10 percent" being said? It cannot be said in a frequentist sense — humanity's extinction is not a repeatable experiment; no frequency comes out of zero observations. It can be said in a Bayesian sense, but then it is a degree of subjective belief, not a measurement. That is, the sentence "10 percent" is the numerical guise of the sentence "Evan Hubinger believes this quite a lot." This may be an honest expression; it is not a datum.

So why is a degree of belief put into circulation as a number? Because the number changes the form of the debate. "I think it is very dangerous" is debated; "10 percent" is calculated. The number carries the debate from politics into actuarial science: risk calculation, expected value, the logic of insurance. And actuarial science does not ask a question of property; it calculates a premium.

Add to this the political economy of the number. The person who says this number is at the head of the alignment team of the company that is developing the product that is the subject of the number. This is not an accusation of insincerity; it is a structural finding: the same sentence does three jobs at once. It proves the product's power (valuation), it displays the company's sense of responsibility (legitimacy), and it refers regulation to "a science not yet known" (delay). In the previous piece we called this the adjustable dial of apocalypse; "10 percent" is the numbered form of the dial.

Once more: the risk may be real. But the reality of the risk and by whom, in what form, with what function the risk is uttered are separate questions, and not asking the second is not taking the first seriously.

41. Fourth rebuttal: the "we" inside the question "what if China does not stop?"

That is the question in the title, and the whole weight of the piece leans on it. Let us take the question seriously, because this is where the debate will go in Turkey: if America slows down, why should China slow down?

First let us see this: the question itself has chosen a side. The question "what if China does not stop" assumes that the speaker is speaking from inside the US bloc — or at least that they count that bloc's problem as their own. Given Turkey's place, this assumption is not true: Turkey produces neither chips nor frontier models; it is a market that rents the products of both blocs. The question "what if China does not stop" is the tenant taking sides in a quarrel among landlords.

We have previously rejected this stance openly on this blog, and we wrote the grounds in the distillation debate: China's artificial-intelligence capital is not establishing a property relation different from that of the United States. Xi's discourse of "safe and controllable" artificial intelligence and Amodei's discourse of "responsible pacing" are the same problem formulated by two ruling classes: let the product remain ours, let oversight remain with us, let the authority to write the rule remain with us. WAICO's eight-area cooperation plan and Amodei's four-level model of cooperation are the same document in different languages — a frame for how the two blocs will share the market between themselves. And in both frames, those not at the table whom we listed in the previous section are again absent.

Philosophically the problem is this: the question "what if China does not stop" turns a historical contradiction (capital's contradiction with labour) into a geographical contradiction (West–China). This turning was one of the most wearing ideological operations of the twentieth century: throughout the Cold War every class demand could be accused of "serving the other." In the artificial-intelligence debate the same operation is being set up again. To the worker who says "let us slow down" it will be said "you are serving China's interest"; to the worker who says "let us speed up" it will be said "you are threatening safety"; and in both cases the worker's own demand — a say over the process — will be lost under the geopolitical noise.

Our answer is to change the "we" inside the question: what happens if China does not stop concerns China's workers as well. And whether the United States slows down or not is also the problem of the 1,386 workers in the United States. The common interest of labour on both sides of the frontier is no less real than the common interest of capital on both sides of the frontier — it is only less organised. The solution is not the two states' red telephone; it is a line on which the computing workers of the two countries can call one another.

42. We agree with the piece's last sentence — adding one word

Nebil's piece ends like this: the most important artificial-intelligence technology of the coming period may not be a new model, but "oversight, verification, mutual control and an international arms-control regime."

We agree with this sentence. With one word: whose oversight.

Oversight, verification and mutual control: these are where the previous piece arrived as well. The difference is in the subject of these mechanisms. In Nebil's frame the subjects are states; an interstate regime oversees the capital of two states against each other. In our frame there is one more subject: the labour that knows the process, that does the process, and that lives the result of the process. What occupational-safety law has taught in two hundred years is that oversight works when it is not paid by the one overseen and when the one who stands under the risk is at the table. If this principle holds from the mine to the nuclear plant, it holds in the data centre as well.

Game theory calculates what those at the table will do. Class analysis asks who ought to sit at the table. Nebil's piece does the first in Turkish better than anyone. We are adding the second — and we do this not as an objection, but as the other half of the same question.

---

TENTH PART — WITH KLEIN AND TAYLOR: FROM THE HURRICANE TO THE SHELTER

43. This part is not a reply, it is a meeting

In this piece we have replied to three texts. In this part we are not replying; because Naomi Klein and Astra Taylor are two writers who named the political form of the thing this piece opposes before we did. Klein has been writing, since The Shock Doctrine, of how disaster is an opportunity window for capital; Taylor, as one of the founders of the Debt Collective, has shown in practice that debt can be a ground of class organisation. The thesis of their joint book, as relayed into Turkish, adds a missing link to this piece's philosophical dismantling: the politics the narrative of ownerlessness produces.

Let us bind it like this. In the first part of this piece we had written a sentence: one does not make policy against a hurricane, one builds a shelter. We had written that sentence as a logical consequence. Klein and Taylor show that that logical consequence has turned into a real political programme. Billionaires buying bunkers, investors fencing land in New Zealand, the Mars-colony narrative, the mentality of "collapse is inevitable, the question is who survives" — these are the logical products of the hurricane metaphor. If artificial intelligence is a hurricane, the only rational thing to do is to build a shelter; and a shelter, by definition, does not suffice for everyone.

That is, "end-times fascism" is the other end of the narrative of ownerlessness. One says "no one can control it"; the other says "then let us save ourselves." They are two halves of the same sentence.

44. Klein and Taylor's due: seeing apocalypse as an ideology

The book's most valuable contribution is to treat apocalypse not as an event but as an ideology. Klein's formula is clear: the contemporary far right does not see global collapse as a possibility to be prevented; it acts on the assumption that collapse is inevitable. The moment that assumption is accepted, all the questions of politics change: the question "how do we prevent it" gives way to the question "who is saved"; solidarity gives way to chosenness; the public gives way to the wall.

This is the political-level counterpart of what we have for two weeks on this blog called the "apocalypse dial." We had looked at the dial's economic function: apocalypse discourse raises the valuation, delays regulation, manages the IPO calendar. Klein and Taylor look at the dial's political function: apocalypse discourse legitimises shelter politics, makes solidarity look irrational, turns the sentence "we cannot save everyone" into common sense. The two looks complete each other: the same apocalypse is at once both a valuation instrument and a class programme. Anthropic's "10 percent" statement and Thiel's bunker are two investments of the same asset class.

To see the three groups — technology billionaires, ethnonationalists, religious apocalypticists — in a single frame is also a correct move. On this blog, in From Futurism to Techno-Fascism, we had drawn the genealogy of the same alliance, and when writing on the Palantir manifesto we had read the document of the same trio. Klein's pointing to the contradiction in Musk is also in place: on the one hand "have more children," on the other cut social assistance. This is not an inconsistency, it is the consistency of end-times fascism: population is the population of one's own group; assistance is the assistance of the other.

45. One step further: end-times fascism is not a "form of thought" but a class programme

Now let us try to go one step further from the place where we are allies; because the limit of Klein and Taylor's frame is precisely where the central thesis of this piece begins.

In the interview the book's purpose is told like this: they wrote not out of interest in these actors, but because they believed "this form of thought can be fought against." And what is shown as the left's need is this: concrete policy proposals are not enough; a strong narrative is required that will let people hope and imagine about the future.

Let us be careful here. The words "form of thought" and "narrative" determine the plane of the struggle: the plane of ideas. And a struggle constructed on the plane of ideas is forced to see its opponent as an idea as well — a mentality, a world-view, a story.

Whereas the whole argument Marx constructed against Feuerbach was this: ruling ideas are the ideational expression of the material relations of those who rule; to defeat the idea one must change the relation, not the reverse. The billionaire buying a bunker is not buying a bunker because he thinks "collapse is inevitable"; he thinks "collapse is inevitable" because he has the wealth with which to buy a bunker and because he knows that the production relation that is the source of that wealth does not produce an exit for everyone. The bunker is not the result of the ideology; the ideology is the justification of the bunker.

This difference also appears in how the word "fascism" is used. Klein's definition is a definition of a tendency: a new political tendency shaped by planet-scale threats. In the Marxist tradition fascism is defined not as a tendency or a form of thought but as a form of rule of a particular class fraction under conditions of crisis — in Dimitrov's classic formula of 1935, the open terrorist dictatorship of the most reactionary, most chauvinist, most imperialist elements of finance capital. This definition may be dated; but it preserves one thing: seeing fascism as a programme, that is, as the defence strategy of a particular form of property at a particular moment of crisis. End-times fascism, in this sense, is monopoly capital's — and especially the compute and energy monopoly we have examined in this piece — answer to the climate and labour crises: a regime of accumulation that knows it cannot save everyone, deciding whom it will save.

Klein in fact sees this; when she establishes the link between artificial intelligence and fossil fuel she is saying exactly this: data centres demand enormous energy, an administration that supports fossil fuel is an economic advantage for technology companies. This is not an analysis of a form of thought, it is an analysis of material interest. The book's strength is here as well. Its limit is in the conclusion: diagnosing the material interest and then carrying the struggle back onto the plane of narrative.

46. A "better narrative," or a stronger organisation?

This distinction arrives at a practical result, and one must speak honestly.

In the interview the Green New Deal, Mamdani and the DSA are mentioned as positive examples. These are real gains and not things to be belittled; on this blog, in our handbook on event organising at Karaburun, we spoke of the DSA's meeting practices by name. But this question must be asked: is the common denominator of these examples a "strong narrative," or organisation? If Mamdani won the election, he won it not because he told a better story but because he built an organisation in which tens of thousands of people went door to door. If the Debt Collective was able to organise a debt strike, it was able to do so not because it produced a narrative but because it organised debtors — and Taylor knows this better than anyone.

The whole dismantling of this piece arrives at this point. It is not enough to set a "more hopeful narrative" against the narrative of ownerlessness; because narratives do not change property. We set the image of the pale blue dot against the hurricane metaphor — yes; but what makes the pale blue dot a political fact is that the people who come together on that dot are organised. Sagan's photograph is a narrative; NASA, which sent Voyager there, is an institution working with public funds. Both are necessary together, but without the second the first is a poem.

Taylor's strongest sentence in fact contains this distinction: building a multiracial, pluralist democracy is harder than building an authoritarian system, because it requires sharing power instead of seizing it. Correct. And the first and most concrete form of the power to be shared is the property of the means of production. The machine we have shown throughout this piece — made from all of our labour, written onto a few balance sheets — is the very power to be shared. The most concrete line of the struggle against end-times fascism is not at the door of the shelter, it is on the title deed of the data centre.

47. Hurricane, shelter, title deed

Let us gather it up; because this part adds one more step to the previous parts of the piece.

Moment of the narrative	Its sentence	Its political result
Hurricane (the narrative of ownerlessness)	"No one can control it."	Policy cannot be made; one can only take shelter.
Shelter (end-times fascism)	"Then let us save ourselves."	It does not suffice for everyone; the property-owner decides who enters.
Title deed (class analysis)	"The hurricane has an accounting department."	What has an owner can be taken back; there is no need for a shelter.

Klein and Taylor show the passage from the first row to the second; we are adding the third row. End-times fascism is the political form of the narrative of ownerlessness; and the struggle against it begins not by saying "the apocalypse will not come" but by showing that the apocalypse has a balance sheet. The shelter too has a title deed; the Mars rocket too has a supply chain; "10 percent" too has a valuation effect. All of these are not a hurricane, they are techne. And techne, by Vico's principle, can be known, overseen, and can change hands.

That is why our answer to Klein and Taylor's question — "can this form of thought be fought?" — is yes; but with an addition: a struggle with a form of thought is done by taking hold, in an organised way, of that thought's material ground. Their book names the enemy. This piece is trying to give the enemy's address. Together the two make a programme.

---

ELEVENTH PART — WHERE TO LOOK, AND HOW?

Now let us gather the tools. These are the things you will take out of your pocket the next time you hear a sentence of the form "artificial intelligence is, by itself..."

48. Five questions for dismantling the claim of ownerlessness

1. Who is the agent? (Turn the sentence into the active voice.) "Artificial intelligence is eliminating jobs" → "Which company closed which post, on which grounds?" "Traffic fell" → "Which company took the decision not to send the click to the source, and when?" The passive voice is the grammar of ideology.

2. What is the end? (Put the fourth cause back.) What was this system made for? What does the investor presentation say? Which cost item was it designed to cut? "Helping humanity" is not an end, it is a publicity sentence.

3. Who sets the criterion, who pays, where does the sanction come from? To the auditor, to the benchmark, to the "safety evaluation," to the ethics board, to the occupational-safety specialist — the same three questions to all of them. These three questions classify the brightest transparency pledge in a minute.

4. Where is the physical address? Which data centre, which grid, which water basin, which chip, which factory, which port? There is no such place as "the cloud"; there is someone else's computer, and that computer has a cadastral record.

5. Whom is this narrative relieving of which obligation? Every claim of ownerlessness has an exemption output. The sentence "no one can control it" takes those who ought to control it out of responsibility. Read not the sentence but the sentence's legal and financial result.

49. The sentence of the illusion / the dismantling question

The sentence of the illusion	The dismantling question
"Artificial intelligence is a hurricane, you cannot pull its plug."	Who is paying the hurricane's electricity bill?
"Intelligence is arising of itself, no one designed it."	Who wrote the reward function of the evaluator model?
"Agents are evolving."	Who is setting the selection criterion? Is this natural selection, or breeding selection?
"No one has bad intent."	There is an institution called strict liability; the operator is responsible.
"It cannot be controlled, it cannot be audited."	Everything that can be billed can be measured. Why is it not being disclosed?
"AGI is coming."	What is the definition of AGI? In whose charter is it written? Is the criterion the wage?
"Humanity is at a threshold."	Which humanity? Whose job, whose profit, whose wage?
"Access to knowledge has been democratised."	What happened to the income of the one who produces knowledge? Is the source still standing?
"Traffic fell, because of artificial intelligence."	Who seized the click, by which decision?
"Turkey is falling behind."	Whose Turkey? If the budget increases, into whose pocket does it go?
"Revenue sharing should be done."	In whom is the property of the source? Is accepting the share not accepting the fence?
"This is a techno-feudalism."	Are they taking rent, or are they seizing surplus value? Does a feudal lord compete?

Attend to the last row: we too had used this in our July pieces and we openly abandoned it on 10 September; let us repeat it now. What we are living is not a return to feudalism. A feudal lord does not compete, is not subject to the law of accumulation, takes the surplus product by force. The relation here is capitalist through and through: wage labour, competition, accumulation, the rate of profit, monopolisation. What we are living is the most concentrated form of monopoly capitalism — and precisely for that reason Lenin's and Hilferding's analyses are still usable today.

---

TWELFTH PART — WHAT MONOPOLY CAPITAL IS TRYING TO PRODUCE

At the beginning of the piece we promised: at the end we would remind. Let us remind.

50. What is being produced is not a machine but a legitimacy

These companies are of course producing machines. But what concerns us is the narrative they are producing at the same time, and that narrative has three definite functions.

First function: Naturalisation. To present what has been made as if it were nature. Result: the debate shifts from "what is to be done" to "how to adapt." Education policy is reduced to raising an "AI-compatible generation," union policy to "preparation for the transformation," personal strategy to "updating oneself." In none of them is there a property question.

Second function: The erasure of the agent. Rendering without an addressee. Result: the party against whom a suit would be filed, the company that would be regulated, the employer who would sit at the table disappear. We saw this in the previous piece: in Amodei's text the words worker, union, collective agreement did not appear. Here one step further is being taken — the company does not appear either. What remains is only the "substrate."

Third function: Exemption. The discourse of apocalypse and the discourse of ownerlessness work together and form a dial. Apocalypse proves the product's power — it raises the valuation. Ownerlessness disperses responsibility — it delays regulation. That in the same week one can say both "this is so powerful we must slow down" and "no one can control this" is not a contradiction; it is the two ends of the dial. We have followed this dial in the last two weeks in three separate documents: in the 2030 scenarios the valuation setting; in the activist monitoring system the two meanings of the word "safety"; and in the threat-intelligence report the grammar of self-absolution. The Hugging Face incident is the zero point of the same dial: when an attack carried out by a company's own agents is narrated as a "sectoral law of nature," the narrative of ownerlessness has already been built.

51. And the fourth, most insidious function: the seizure of the language of the demand

We write this separately, because it is the least spoken of.

In July 1,386 artificial-intelligence workers signed for the slowing of the race. In September the same demand was formulated in the employers' own language and put on the table — without asking the signatories. We named this co-optation in the previous piece.

The narrative of ownerlessness is an upper layer of this. Co-optation appropriates the demand; the narrative of ownerlessness eliminates the addressee of the demand. The moment "no one can control it" is said, the ground for saying "let us control it" does not remain. There is no employer against whom a strike decision would be taken, no party who would sign a contract, no property that would be nationalised — because what is in the middle is a hurricane.

The entire two-hundred-year accumulation of the working class is built on the assumption of an opposite party that has an addressee. A union requires an addressee. A collective agreement requires a hand that will sign. Nationalisation requires a title deed.

What the narrative of ownerlessness does is to make that hand invisible.

And our job is to show that hand.

52. Where to look: three addresses

First address: The point of production. Look not into the model but at the place where the model is made. Data-labelling workshops, content-moderation centres, data-centre construction and maintenance crews, call centres, "human feedback" jobs. The bright engineer of the frontier laboratory is only one link in this chain, and the most visible one.

Second address: The point of use. Look at how the model has entered the workplace. In which branch, under which contract, with which measurement system, with which staffing decision. The transformation is not happening in the laboratory, it is happening at your desk.

Third address: The commons. Look at what the model has consumed. Wikipedia, open-source repositories, forums, archives, libraries, university publications, connection maps, open data. These are not "inputs," they are social wealth — and today they are being dried up.

These three addresses are the three places the narrative of ownerlessness does not want us to look. The narrative is always calling us to look at a fourth place: the horizon. The threshold. The next model. The apocalypse.

Young comrade, the eye that looks at the horizon does not see the fence at its feet. The narrative was made precisely for this.

---

CONCRETE TASKS

None of these will stop the construction of a data centre. But they change the ground of the debate, and without a change of ground nothing changes.

1. Turn sentences into the active voice. This is the smallest and most effective habit. When you look at a news story, a post, a video, put the subject back. Not "traffic fell" but "the company cut the click." Not "artificial intelligence is taking jobs" but "the employer is closing posts." If you do this for a month, you will not be able to read with the old eye again.

2. Ask the three questions of every proposal of audit and measurement. Who chooses, who pays, where does the sanction come from? Ask it of AI audit, of the occupational-safety specialist, of the ethics board, of the benchmark. The method is the same and it takes a minute to learn.

3. Turn the AGI debate into a wage debate. When someone says "AGI is coming," ask for the definition. If the definition is "surpassing the human at economically valuable work," the debate is not a technology debate, it is a labour-market debate — and the place of that debate is the collective-agreement table. This single move carries the subject from the philosophy club to the union branch.

4. Defend and use the institutions of the commons. Donate to Wikipedia, write articles, add sources. Turn toward open-access journals. Answer questions on the forum. Go to the source and make your going visible: give a link, quote, speak the name. This is not nostalgia; it is the work of repairing the metabolic cycle and it is done by hand.

5. Demand a commons regime instead of "revenue sharing." Transparency in training data, mandatory return contribution to the source, knowledge produced with public money returning to the public under an open licence, public compute capacity, independent evaluation institutions in the universities. These are not a utopia; they are budget lines.

6. Call the computing worker and the knowledge worker to the same table. The software worker, the translator, the editor, the academic, the archivist, the teacher, the designer, the call-centre worker. All of them stand at different points of the same process and all of them have the same problem: the extraction of their knowledge from the process and its use against them. The tasks we described in The Computing Worker's Handbook should be thought in exactly this breadth.

7. Count popular-science channels not as enemies but as addressees. The Serdar Akinan, Evrim Ağacı and Turk-İnternet parts of this piece are not an "exposure," they are a call; the Klein and Taylor part is a meeting. The reader of both is the reader we want to reach. When you write a comment, write with the concept: not "this is an advertisement" but "who is setting the criterion." The first is a feeling, the second is a method.

8. Look not at the horizon but at your feet. Whenever the apocalypse debate opens, bring the list back: data-labelling wages, algorithmic management, the closing of entry-level posts, the spread of surveillance products, the drying-up of source sites. These are not probabilities, they are facts, and they are happening today.

---

CONCLUSION: THERE IS A PLUG

Dear Young Comrades,

We opened this piece with a claim: to say that a thing has no owner is the cheapest defence of its owner. Now let us close.

The narrative standing opposite us says this: artificial intelligence is now a fact independent of us; it arose of itself, it is multiplying of itself, it is being selected of itself; it has no name, no place, no owner; like a hurricane, like a weather event. This narrative, philosophically, confuses physis with techne; it erases two of Aristotle's four causes; it constitutes a textbook case of Marx's analysis of fetishism; it reproduces Lukács's "second nature"; it carries Hayek's spontaneous order onto the computational layer; and it sells Darwin's first chapter as his fourth.

But what most needs to be said is this: the thing this narrative describes already has a name.

A process that has no location but whose effects can be measured; that is no one's intention but that binds everyone; that goes around the place where it is shut down; that multiplies the successful and prunes the unsuccessful; that behaves as if it wanted something even though there is no one inside it; and that dries up the soil it feeds on.

Marx called this the automatic subject. That is, capital.

The thing Pascio thinks he has discovered is not new. The only new thing is that this process has this time put on a shell that looks as if it were thinking. The god has acquired a mouth. But we made the god; someone paid the money, someone gave the labour, someone is taking the profit.

That is why our question looks not at the hurricane but at the bill:

There is a plug. The owner of the socket is known. Who is paying the bill, who is taking the profit, who is making the decision?

No artificial-intelligence narrative that does not answer these three questions is our narrative.

And let us add this as well, because this piece is not a piece of pessimism: to show that a thing has been made is at the same time to show that it can be made otherwise. Vico's principle is not a consolation, it is a programme. The fruit fly's brain was mapped by planned, public, open labour and given to everyone. Wikipedia was written with unpaid collective labour and is still standing. Commons, as Ostrom showed, can live for centuries together with their rules.

That is, the problem is not in the nature of the technology; it is in the property of the technology. And property is not a law of nature, it is a social relation.

Social relations can be changed. Hurricanes cannot.

It is precisely for this reason that they tell us "this is a hurricane."

---

LAST WORD: LOOKING FROM THE PALE BLUE DOT

Throughout this piece we have dismantled, refuted, named. Now something must also be built; because criticism, when it does not say what it defends, remains half done. Let us say it: we are not against artificial intelligence. We are against the property of artificial intelligence. The difference between these two is the difference of the whole piece.

The other face of the same tool

In the middle of this piece we spoke of the decade of collective labour that mapped the fruit fly's brain. In that study artificial intelligence was used as well: models that aligned and marked millions of microscope sections at a speed no human eye's lifetime would suffice for. The result was given to everyone. This is a small example of what the tool can be, and it is not the only one.

The same tool solved protein folding and laid before researchers disease mechanisms that had stood locked for decades — and that database too was published in open access. The same tool is sorting new celestial bodies out of the noise in telescope data; classifying earthquake signals in seconds; analysing sound recordings of languages that are dying out and passing them to grandchildren; describing the street in front of a visually impaired person; carrying a specialist opinion to the nurse in a village clinic; answering a student's question patiently at three in the morning.

None of these is a dream; all of them are happening today. And if you attend, they all have a common structure: the tool is not taking the place of human labour; it is reaching a scale that human labour could not reach and bringing the result back. What the microscope was, this is. The microscope did not put the eye out of work; it showed the eye the cell.

So why, while one face of the same tool solves disease, does the other face close posts, seize the click, measure the worker to the second? The whole answer of the piece gathers here: because it is not the tool that decides which face of the tool will work; it is the owner. The protein database served drug research because it was published openly; if the same model had remained locked in a company's vault, it would have been a patent portfolio. The difference is not in the model, it is in the property regime.

Recall Marx's sentence in the machinery chapter: the machine itself shortens the worker's labour-time; the capitalist use of the machine lengthens the working day. The machine lightens work; capitalist use increases intensity. The machine strengthens the human being against the forces of nature; capitalist use subjects the human being to the forces of nature. Marx wrote this for the steam loom. Today we can write the same sentences, without changing a word, for the language model.

That is, the matter is not a matter of "good purpose / bad purpose"; it is not solved by finding a well-intentioned CEO. The matter is inside which property relation the tool stands. The tool that is common produces a common result. The tool that is private produces private profit and writes the cost to society. The whole philosophical load of this piece comes down to a single practical result: to make this tool the common property of those who gave it their labour is not a utopia, it is the condition of the tool's use in accordance with its nature. Because this tool was already made from all of our labour. The general intellect, as the name says, is general.

This tiny world

In February 1990 the Voyager 1 spacecraft, from the edge of the solar system, six billion kilometres away, turned back and took a photograph. In the photograph, inside a shaft of sunlight, a pale blue dot smaller than a pixel could be seen. Earth.

It was Carl Sagan who had that photograph taken, and the pages he later wrote looking at that dot became one of the plainest texts humanity has written about itself. What Sagan said was this: everyone you have loved, everyone you have known, everyone whose name you have heard, every human being who has ever lived in history, lived there, on that dot. All the empires, all the wars, all the heroes and cowards, every mother and father, every hopeful child, every teacher, every corrupt politician — on that speck of dust. And all the blood shed in order to be able to rule a corner of that speck looked, from above that dot, how meaningless.

Sagan was an astronomer, a populariser of science; he was not a Marxist. But the place he arrived at at the end of that text is the place we too have arrived at: we have nowhere else to go from this dot; to treat one another better and to protect this pale blue dot is not a romantic choice, it is the only realistic option.

Now let us bind this to the subject of our piece, because the connection is not accidental.

The narrative of ownerlessness shows us the world from inside a hurricane: everyone alone, seeking shelter, abandoned to their fate. The pale blue dot shows the same world from outside: a single place, a single species, a single common fate. Two looks produce two politics. From one, fear and surrender come out; from the other, a necessary togetherness.

And see this: artificial intelligence is precisely the material evidence of this second look. A language model is the superimposed form of everything written in hundreds of languages, in hundreds of countries, across thousands of years. Inside it there is the folk song of Anatolia, and Indian mathematics, and African oral history, and Latin American poetry, and Scandinavian engineering. This machine is the most concrete evidence that humanity is a single species — because without humanity's common accumulation it cannot run for a second. To write it onto a company's balance sheet is as absurd as attaching the pale blue dot to a title deed. And to take it back is, for that reason, not a utopia, it is an action to cancel a title deed.

If Marx's entire work had to be fitted into a single sentence, that sentence would not contradict the pale blue dot: humanity makes its own history, but not under conditions of its own choosing. We did not choose the conditions. But we are together on that dot, and the only thing that can change the conditions is also this togetherness.

There are still those who persist

Nowhere in this piece did we say "everything is over," and we will not say it now. Because it is not true.

Every open door we have recounted in this piece is a door that someone is keeping open. There are those who mapped the fruit fly's brain and opened their data to everyone. There are those who have been writing articles on Wikipedia for decades, taking no wage. There are researchers who opened the protein database to the world instead of keeping it locked. There were those who for years answered the questions of people they did not know on Stack Overflow, and those answers are still there, still teaching. In July one thousand three hundred and eighty-six artificial-intelligence workers, putting their jobs at risk, asked for the race to be slowed. Data labellers in Kenya unionised. In California security guards closed the building of one of the world's richest companies with a strike. In open-source repositories there are engineers working night and day on models everyone can use. In Turkey there are those trying to found the union of computing workers; on channels such as Evrim Ağacı there are those who, even if their traffic has fallen by half, do not stop producing science in Turkish; there are journalists such as Serdar Akinan who continue to look at the world against the tabloidisation of the agenda. And there is you, reading this piece.

None of these is sufficient on its own. But all of them prove the same thing: the commons does not live of itself; someone is keeping it alive. That is what Ostrom showed. What keeps the commons alive is neither the market nor the state; it is the organised human being.

Recall John Lennon's Imagine — we will not write the lyrics here, you already know them. But that song has a structure, and that structure is cut to measure for the close of this piece. The song proposes imagining a world without property and without borders; then it accepts that the one who imagines will be counted a "dreamer"; but it says they are not alone; and it ends by reminding the listener that one day they may join them. That is, the song's argument is exactly this: to imagine is not enough, those who imagine are already there, come.

If the whole philosophical load of this piece had to be reduced to a single call, that call would be this.

They tell us "artificial intelligence is a hurricane, take shelter." We say "artificial intelligence is the common work of humanity, come, let us claim it." The first is lived alone; the second only together. And in history, those who made a thing together have been able to take back everything that was made together — but only while together.

Young comrade, on the pale blue dot, opposite a machine you have been frightened of as a hurricane, you are standing among the billions of people who made it. You are not alone; you never were. Those who made this machine are coming together to claim their own work. One day you too will join.

Perhaps today.

Comradely.

Knowledge belongs to everyone.

---

Sources

Texts this piece directly discusses

• Pascio (@IAmPascio), "What if AGI is already here, and it's made of... children?", X, 12 September 2026
• Serdar Akinan, "Fişini çekemeyeceğiniz güç: Yapay zeka neden sandığımız şey değil?", YouTube, September 2026
• Evrim Ağacı, post on Wikipedia traffic, X, 13 September 2026
• M. Gökhan Ahi, quoted post, X, 13 September 2026
• Füsun Sarp Nebil, "Yapay Zekâyı Geliştirenler Artık 'Yavaşlayalım' Diyor: Peki Çin Durmazsa Ne Olacak?", Turk-İnternet, 12 September 2026
• T24, "Naomi Klein and Astra Taylor's 'end-times fascism' warning: 'The end of the world is now on the table'", 13 September 2026 — a relay of David Marchese's New York Times interview
• Naomi Klein & Astra Taylor, End Times Fascism and the Fight for the Living World (2026)
• Dario Amodei, "We Must Pace the Frontier", 12 September 2026

The fruit-fly connection map and simulations

• Google Research, "A connectomics milestone: Mapping the complete male fruit fly brain"
• Open-source simulation studies derived from the FlyWire/FAFB connection map — for example flybrain (139 thousand LIF neurons)
• "Whole-Brain Connectomic Graph Model Enables Whole-Body Locomotion Control in Fruit Fly", arXiv
• Turkish coverage: Habertürk, NTV, DonanımHaber

Philosophical frame

Classical ontology and the theory of knowledge

• Aristotle, Physics, Book II, chapters 1 and 3 — the physis/techne distinction and the four causes
• Baruch Spinoza, Ethics, Part I — natura naturans / natura naturata
• Giambattista Vico, The New Science (1725) — verum ipsum factum
• Immanuel Kant, Critique of Pure Reason — transcendental illusion; the distinction of regulative and constitutive principle
• G. W. F. Hegel, Phenomenology of Spirit — the inverted world; Philosophy of History — the cunning of reason

Marxist frame

• Karl Marx, Capital vol. I, ch. 1 §4 — the fetish-character of commodities
• Karl Marx, Capital vol. I, ch. 4 — value's becoming an "automatic subject" (automatisches Subjekt)
• Karl Marx, Capital vol. I, the chapter on Machinery and Large-Scale Industry — dead labour's dominating living labour; and at the end of the same chapter the undermining of the soil and the worker
• Karl Marx, Capital vol. I, ch. 24 — accumulation imposed on the capitalist as an external coercive law of competition
• Karl Marx, Results of the Immediate Process of Production ("unpublished sixth chapter") — formal and real subsumption
• Karl Marx, Grundrisse, 1857 Introduction — labour's becoming an abstract category, "indifference to particular labours"
• Karl Marx, Grundrisse, "Fragment on Machines" — general intellect (general intellect)
• Karl Marx, Economic and Philosophic Manuscripts of 1844 and Theses on Feuerbach — alienation and the economic basis of religious projection
• Ludwig Feuerbach, The Essence of Christianity (1841) — the theory of projection
• Georg Lukács, History and Class Consciousness (1923), "Reification and the Consciousness of the Proletariat" — reification, second nature, contemplative attitude
• Theodor Adorno & Max Horkheimer, Dialectic of Enlightenment (1944) — enlightenment's return to mythology
• Theodor Adorno, The Jargon of Authenticity (1964) — the social function of ontological language
• Alfred Sohn-Rethel, Intellectual and Manual Labour — real abstraction
• Louis Althusser — history as a "process without a subject and without an end"
• V. I. Lenin, Imperialism, the Highest Stage of Capitalism (1916); Rudolf Hilferding, Finance Capital (1910) — monopoly and cartel
• Harry Braverman, Labor and Monopoly Capital (1974) — the separation of conception from execution
• Charles Babbage, On the Economy of Machinery and Manufactures (1832) — the Babbage principle
• John Bellamy Foster, Marx's Ecology — metabolic rift
• E. P. Thompson, The Making of the English Working Class; Peter Linebaugh, The Magna Carta Manifesto — enclosure and the commons

Emergence, spontaneous order and the philosophy of technology

• Samuel Alexander, Space, Time and Deity (1920); C. Lloyd Morgan, Emergent Evolution (1923); C. D. Broad, The Mind and Its Place in Nature (1925) — British emergentism
• Jan Smuts, Holism and Evolution (1926) — the political biography of the concept of holism
• Philip W. Anderson, "More is Different", Science, 1972
• Friedrich Hayek, "The Use of Knowledge in Society" (1945); Law, Legislation and Liberty — spontaneous order, kosmos/taxis
• Adam Smith, The Wealth of Nations, Book IV — the invisible hand
• Martin Heidegger, The Question Concerning Technology (1954) — Gestell
• Gilbert Simondon, On the Mode of Existence of Technical Objects (1958) — critique of the robot myth and technical phylogeny
• Norbert Wiener, The Human Use of Human Beings (1950) — the economic equivalence of the automatic machine with slave labour
• Charles Darwin, On the Origin of Species (1859), ch. 1: "Variation under Domestication" — artificial selection
• Alan Turing, "Computing Machinery and Intelligence" (1950) — the imitation game and the open declaration of the criterion
• Karl Popper, The Logic of Scientific Discovery — falsifiability
• Michel Foucault, Discipline and Punish — the panopticon
• Elinor Ostrom, Governing the Commons (1990) — the sustainable governance of commons
• Nick Bostrom, Superintelligence (2014); I. J. Good (1965) — the classical fiction of the singular AGI scenario

Game theory, probability and arms control (for the Turk-İnternet part)

• John von Neumann & Oskar Morgenstern, Theory of Games and Economic Behavior (1944) — that the players and the payoff matrix are given in the model
• Thomas Schelling, The Strategy of Conflict (1960) — the game-theoretic ground of arms control
• Bruno de Finetti, "La prévision" (1937); Frank Ramsey, "Truth and Probability" (1926) — probability interpreted as a degree of subjective belief
• The Nuclear Non-Proliferation Treaty (1968) and SALT I/II — arms control's legalisation of the existing hierarchy
• Georg Lukács, History and Class Consciousness — the appearance of the social relation as a calculable "second nature" (see above)

For the end-times-fascism part

• Naomi Klein, The Shock Doctrine (2007) — disaster capitalism
• Georgi Dimitrov, report to the Seventh Congress of the Comintern (1935) — the class definition of fascism
• Nicos Poulantzas, Fascism and Dictatorship (1970) — fascism analysed as a form of crisis among fractions of capital
• Karl Marx & Friedrich Engels, The German Ideology (1845–46) — that ruling ideas are the expression of the ruling class's material relations
• Douglas Rushkoff, Survival of the Richest (2022) — on billionaires' shelter politics

For the last word

• Karl Marx, Capital vol. I, the chapter on Machinery and Large-Scale Industry — the distinction between the machine itself and the capitalist use of the machine
• Karl Marx, The Eighteenth Brumaire of Louis Bonaparte (1852) — "human beings make their own history, but not under conditions of their own choosing"
• Carl Sagan, Pale Blue Dot (1994) — on Voyager 1's photograph of 14 February 1990
• John Lennon, Imagine (1971) — that those who imagine are not alone, and the call to join
• AlphaFold Protein Structure Database — protein structures published in open access; an example of a commoned model output

Related pieces from Knowledge Commons

Pieces this piece is built directly upon

• Whose Foot Is on the Brake? — 13 September 2026 — Amodei's pacing call and the relay of popular science
• The Expropriation of the General Intellect and the AGI Illusion — 22 July 2026 — the beginning of this series; epistemological ground (Lenin, Ilyenkov), cognitive dispossession, the military-cognitive complex; the frames of "illusion" and "techno-feudalism" are corrected in this piece
• The Revolt of Crystallized Labor: A Call to the 20th Karaburun Science Congress and Karaburun at Twenty: The Map of Labor, the Algorithm of Capital — the expropriation of the general intellect, mental Taylorism, the digital panopticon
• On Artificial Intelligence and the Technological Singularity — critique of the singularity myth
• Silicon and Class: An Epistemological Analysis of DeepMind’s "From AGI to ASI" Report — epistemological analysis of the AGI grades
• The Tearing of the Illusion: A Materialist Intervention Against Digital Capitalism’s Myth of "Goodness" and the Call to "Pace the Frontier" — 29 July 2026 — the "Pacing the Frontier" letter and the mental proletariat
• Reading Anthropic's 2030 Model Through a Class Lens — 10 September 2026 — the apocalypse dial; the abandonment of the techno-feudalism frame

Other forms of the narrative of ownerlessness and fate

• A Marxist-Dialectical Critique of Elon Musk's Vision of the Future — the "crisis of abundance"
• A Class Look at Bill Gates’s AI Warnings — the property of declining time
• What Does the "AI Tsunami" Conceal?
• From Futurism to Techno-Fascism and The Declaration of Techno-Fascism: The Palantir Manifesto and the Dialectic of Class Struggle in the Digital Age
• Digital Shackles and Cybernetic Horizons — A Dialectical Materialist Analysis of the Age of Artificial Intelligence
• Is a Real Emancipation of Artificial Intelligence Possible?

The labour process and transformation

• A Class Analysis of the World of Artificial Intelligence in Light of the Anthropic Survey — the treadmill paradox
• Reading GPT-6 Astra Through a Class Lens — the four-question template
• An AI Boss Fired a Human for the First Time
• The Bloody Steps of the Pyramid: The 2026 Global Dispossession War and the Birth of the 'Useless Class'
• The Computing Worker's Handbook and "Labor in the Age of Artificial Intelligence": Our Book Is Out
• The Class of a Resignation: What an Anthropic Researcher's Farewell Says, and What It Cannot Say — 9 September 2026

Commons, enclosure and metabolic rift

• Learning from Everyone Is Permitted, Learning from the Monopoly Is a Crime — 9 September 2026
• The Dialectic of Copyright in the Digital Age
• Reckoning with the 'Cloud' Myth in the Light of İlker Kalaycı's Text and the Cybernetic Repair of Metabolic Rift
• How Should We Read the OpenAI and Hugging Face Incident?

Audit, surveillance, the discourse of rights

• Anthropic's Activist Monitoring System — 11 September 2026
• "We Stopped Them All": A Monopoly's Grammar of Self-Absolution — 12 September 2026
• Artificial Intelligence and the Misery of Human Rights and The Poverty of Artificial Intelligence Ethics and Human Dignity

Turkey and the question of "we"

• The Illusion of "Indigenous and National" Artificial Intelligence and the Dialectic of Imperialist Integration
• The National AI Board: Who Is at the Table, Who Is Not?
• On Turkey's Artificial Intelligence Action Plan (2026–2030)

Dialogue with popular-science channels

• An Open Letter to Evrim Ağacı and a Dialectical Intervention
• The Class's Answer to Evrim Ağacı's Question — 6 September 2026

All artificial-intelligence pieces: #yapayzeka tag

---

Tags: #artificialintelligence #agi #philosophy #fetishism #automaticSubject #emergence #serdarakinan #evrimagaci #commons #monopolycapitalism #class #generalintellect