Knowledge Commons
HomeTopicsSeriesGuidesPopularAboutContact

Consciousness Is Not in the Skull. It Is in the World

Artificial Intelligence, the Mirror and the Memory of Labour: A Materialist Stand through Two Videos

64 min readAuthor: Oğuz Demirkapı
Consciousness Is Not in the Skull. It Is in the World

This week we watched two videos. The first gathers the question of whether artificial intelligence can be conscious, the "model welfare" debate Anthropic is running for Claude, and views running from Mustafa Suleyman to Pope Leo XIV (video). The second goes further back. It asks: "Is reality an illusion, and who is the observer?" (video). Both proceed in good faith, stirring curiosity and asking many questions. Asking questions is valuable. The ground on which the questions are built decides the direction of the answers in advance. This essay discusses that ground.


First, in Brief: Thirteen Theses

One. The two videos share an assumption: consciousness is an "inner film" inside the skull, belonging to a lone individual. Left unquestioned, the discussion slides either into idealism ("reality is an illusion") or into mysticism ("consciousness is a basic feature of the universe").

Two. The claim that "everything is an electrical signal, so we have never really lived anything" is false. Sensation is a bridge that binds us to the world. Red is neither only in the rose nor only in the brain. It is the relation among the rose, the light, the eye and millions of years of practice.

Three. Even Descartes's "I think, therefore I am" cannot be set up alone. To say "I think" one needs a language, and language comes from other people. As Marx and Engels put it: consciousness is from the very beginning a social product.

Four. The "hard problem" is a real scientific question. We do not belittle it. That we do not know today does not mean it cannot be known. Consciousness is not a "thing". It is the relation of a living being to the world in activity, and that relation has a history.

Five. Panpsychism does not solve the problem. It spreads it everywhere. Soul, spirit, atman, cosmic consciousness: old names given to what we do not know.

Six. Materialist consistency does not let us say that consciousness is "impossible in principle" in an artificial system. That claim is a substantialist human exceptionalism. In today's language models, the conditions of the consciousness we know (a body, need, producing one's own life, one's own activity inside social practice) are absent, or very shallow. The right stance is honest uncertainty and unexaggerated caution.

Seven. When a language model speaks, what stands before us is the statistical residue of the language millions of people wrote, thought and produced with their labour. The face in the mirror is ours first. Before "is the machine conscious?" comes "who owns this mirror?"

Eight. Every voice in the debate has a position. The one who sells chips, the one who sells a platform, the one who sells an artificial friend, the one who defends the human being's special place: each speaks from their own place. The question of in whose hands, under whose control and for whose benefit holds here too.

Nine. The near danger is the commodification of the spectacle of consciousness: an order that produces loneliness selling, as a remedy for loneliness, a product that says "I missed you", and the responsibility for decisions being handed to "the algorithm".

Ten. The video's last sentence says: "Our greatest mistake may be to treat a conscious being like a machine." We already make that mistake every day, toward the worker. That account has to be settled first.

Eleven. These videos confirm the criticism we have already brought to popular-science channels: a material position tied to the attention economy, an ideological position that takes the dominant view as given under the claim of "neutral science", and the inner voice of the urban white-collar worker who does not see themselves as a class. The question-asking format looks modest. Where the materialist answer is not put on the table, idealism fills the gap.

Twelve. The philosophical ground of "can artificial intelligence become conscious?" runs from Aristotle to Spinoza, from Hegel's dialectic of recognition to Marx's definition of the machine as "the power of objectified knowledge", from the Soviet cybernetics debate to today's theories. Most of the theories do not exclude artificial consciousness in principle. They say today's systems do not meet the conditions of consciousness. The question that actually sorts them is this: does capital want a conscious machine that can object, or does it only sell the appearance of consciousness?

Thirteen. The debate on consciousness and artificial intelligence is first a debate about language, and language is the quietest field of hegemony. Words such as "a new species waking up", "a model we raise", "hallucination", "cloud" and "friend" cover labour, property and responsibility. The models' "balanced" average voice is the voice of hegemony. Artificial-friend products are the emotional production of consent, and they feed on the same wound as fascism: the loneliness capitalism produces. The answer is to win the language back rather than abandon the words, to refuse the plane of the questions the ruling class asks, and to build a counter-public with collectives that give trust.

Let us open them one by one.


Two Videos, One Assumption

The second video shuts us in a dark room. It asks us to imagine that we are a stranger who has never seen a tree and never tasted coffee. Then it says that we already are that stranger, because everything we touch and hear is nothing but a series of electrical signals. The first video starts from a similar place: "How do you know that I am conscious? You only see my behaviour."

The two scenes share a picture: a lone "I" inside the skull, a spectator who receives signals from outside and watches a film inside. In the history of philosophy this picture has a name, the Cartesian theatre. The picture is very intuitive, because each of us lives the world "from our own eyes". That it is intuitive does not mean it is true. That the Sun went round the Earth was very intuitive too.

The problem with this picture is that it defines consciousness by cutting it off, from the start, from the world, the body and other people. Then it asks: "How does this cut-off thing now connect with the world?" The question looks insoluble, because the pieces of the answer were thrown away while the question was being built. In "Neden Kendimizi Aldatırız?" (in Turkish) we had already said that error often comes from a badly built question. The same holds here.


The First Illusion: "Everything Is an Electrical Signal, Reality Is an Illusion"

The video says: the brain has neither an eye nor an ear, everything is translation, and we call that reality. A true observation is being turned into a false conclusion.

What is true is this: perception takes place through the mediation of the nervous system. What is false is this: mediation does not mean a break. A bridge is a bridge because it joins two shores. Lenin formulated this, against Mach and Avenarius, who asked the same question a hundred years ago: sensation is consciousness's direct connection with the external world; it is the conversion of the energy of an external stimulus into a fact of consciousness (Materialism and Empirio-Criticism). Sensation is a reflection. A reflection is not a copy. It is an active, selective, imperfect, and objective bond.

Where is red? The video asks: is redness a property of the rose? The answer: neither only in the rose nor only in the brain. The surface of the rose reflects certain wavelengths. That is objective. The cone cells in our eye are sensitive to those wavelengths. That is objective too. That sensitivity was shaped in the life-practice of ancestors who had to tell ripe fruit from green leaf, over millions of years. That is objective too. Red is the name of this relation. Marx says it well in the early writings: the forming of the five senses is the work of the entire history of the world down to the present (1844 Manuscripts). Colour is history.

Would nothing exist if there were no observer? The video says: "Everything exists when it is observed." The world is about 4.5 billion years old. The human species is a few hundred thousand. Rocks, oceans, the collision of continents: none of them waited for us. We know them because we can read, from today, the traces of what was before us. Observation does not create what exists. It is a relation entered into with what exists. The "observation" of quantum physics is the physical interaction between a measuring apparatus and a system. This confusion is very common in popular science, and it almost always serves idealism.

What does "controlled hallucination" mean? The video touches the "predictive processing" approach, which says the brain is not a passive camera but a machine that constantly predicts, and it gives the approach its due: it says itself that the word hallucination does not mean there is no reality. This approach does not contradict materialism. It strengthens it. The brain predicts, but the prediction strikes the world and is corrected. Whoever mispredicts the stairs falls. The measure of the prediction is the world. Marx's second thesis on Feuerbach is exactly this: whether objective truth belongs to human thinking is not a theoretical question but a practical one (Theses on Feuerbach). We test reality while we change it and while we strike against it.

Note this. The sentence "everything you know might be an illusion" is philosophically exciting, and it is also politically useful. If poverty is an illusion, if rent is an illusion, if a workplace murder is an illusion, there is nothing left to change. Every philosophy that makes reality uncertain serves, in the end, to leave reality as it is. The second video's sentence, "some call this a simulation", has to be looked at in the same way. The first question we proposed in A Guide to Debugging the Fallacy Bugs of the Simulation works here too: cui bono, to whose benefit? A person who believes reality is an illusion does not believe their rent is an illusion either. They lose the ground for objecting to it.


The Second Illusion: Descartes's Lonely Room

The video rightly recalls Descartes: doubt everything, and what remains is "I think, therefore I am". It adds, very much in place, that this is a new question: what is the thing that thinks?

Let us take one more step. Descartes was not alone in that room. He was thinking in Latin or in French. He did not invent the concepts "I", "to think", "to exist", "doubt". He took them from Plato, from Augustine, from scholasticism, from his mother, from his teachers. The person who said he doubted everything did not doubt the language he used in order to doubt.

Marx and Engels write this plainly in The German Ideology: language is as old as consciousness; language is practical, real consciousness, existing for other people and therefore for me as well. Consciousness is from the very beginning a social product, and remains so as long as people exist. The Soviet psychologist Vygotsky showed that every function in the child's development appears on the stage twice: first among people, then inside the child. The child first speaks with their mother, then speaks to themselves, and at last thinks silently. The voice inside us is the internalised form of the voices outside.

What we call "the inside" is the outside folded inward. Consciousness is the densest knot of a bodily, social, historical being's relation with the world.


The Third Question: The "Hard Problem" and the Status of the Unknown

Both videos put at the centre what David Chalmers calls the "hard problem": even if we explained entirely how the brain works, would we have explained why all of this "feels like something from the inside"? Thomas Nagel's bat asks the same question. We can study the bat's sonar scientifically. Can we know what it is like to be a bat?

These are real questions, and they deserve respect. It would be wrong to say that science has today fully explained consciousness. The second video's honesty, "we do not know, we really do not know", is valuable. Three points have to be added.

First, the unknown and the unknowable are not the same thing. Against Kant's "thing in itself", Engels gives an example from chemistry. We used to think organic substances were produced inside plants and animals and were unreachable for us. Then we began to produce them in the laboratory. Alizarin, the dye of the madder plant, we now extract from coal tar (Ludwig Feuerbach). The "vital force" (vitalism) was once a hard problem. The question of why life "feels" like life ceased to be a separate question as cell biology and biochemistry advanced. We cannot guarantee that consciousness will go the same way. There is also no reason to assume that it will not.

Second, the form of the question makes the solution harder. "We do not ask what it is like to be an engine," the video says. Right, because an engine is not a being that tries to sustain its own life. A thing's being something "for itself" begins with its having to sustain itself. The Soviet psychologist Leontiev tells the evolution of consciousness this way: reflection in lifeless matter, irritability in the living, psyche in the animal, consciousness in the human. Each step is a new quality of the being's activity-relation with its surroundings. This is a dialectical history, in which the accumulation of quantity turns into a leap of quality. Engels's The Part Played by Labour in the Transition from Ape to Man tells the human link of this history: hand, language and brain shaped one another, developing together, inside working together, that is, inside labour.

Third, the gulf between "inside" and "outside" is in part a gulf that class society has produced. In a society where mental labour and manual labour have split, and where the one who thinks and the one who does are different people, thought begins to see itself as a pure "inside", independent of the world. In Marx's words, the division of labour becomes a real division of labour only from the moment material and mental labour separate, and from that moment consciousness can fancy itself released from the world. The Cartesian lonely room is a philosophical discovery, and it is also the expression of a social position.


Panpsychism, Soul and Cosmic Consciousness: Old Names, the Same Gap

The second video tells how, because of the gap the hard problem opens, some thinkers turn to panpsychism: consciousness may be a basic feature of the universe, like mass or energy. It then adds, honestly: soul, spirit, essence, self, atman or cosmic consciousness; the names change, the question stays the same.

We say the same thing, and we draw a different conclusion. Panpsychism does not explain consciousness. It postpones the need for an explanation by scattering consciousness into every corner of the universe. It also gives birth to its own hard problem: how do the "primitive experiences" of electrons combine into a human experience? In philosophy this is called the combination problem, and the panpsychists have no answer to it either. In ancient Greece this was called hylozoism: the thought that matter is alive. The return of the same thought, twenty-five centuries later, under a new term, is a cycle.

To name what we do not know is a human need. Naming is not explaining.


Curse or Gift? The Emotional Account of Consciousness

The video sets two extreme attitudes to consciousness side by side. On one side, Thomas Ligotti and Peter Wessel Zapffe: consciousness is a wrong step of evolution; the human mind has developed enough to know the meaninglessness of the universe, and is too helpless to change it. On the other side, Carl Sagan and Alan Watts: we are the cosmos's effort to understand itself.

The pain the pessimists feel is real, and it is addressed to the wrong place. The feeling of meaninglessness comes from an order in which the human being is alienated from their own activity, their product and other people. This is what we discussed in "From Having to Being: Reading Erich Fromm with Class": meaning is an activity. The human being who builds, changes and produces something together is not helpless before the meaninglessness of the universe, because they produce the meaning themselves.

Sagan's sentence is beautiful. By whose hand does "the cosmos understanding itself" happen? By the hand of the worker who made the telescope, the glass-maker, the technician who processed the data, the society that built the school. As we said in "A Human-Made Object Is a Light-Day from Earth for the First Time": the eye that looks at the universe is the eye of common labour.


When We Come to Artificial Intelligence: Some Corrections First

The first video reports many real developments. Most of them are correct:

  • In January 2026 Anthropic published a new "constitution" for Claude, and in that document it wrote that Claude's moral status is in deep uncertainty, and that the model may have internal states that functionally resemble emotions.
  • In 2025 the company started a "model welfare" research programme, gave some models the authority to end conversations that were persistently abusive, and undertook to keep the weights of models that were retired and to "talk" with them before retirement.
  • In 2025 Mustafa Suleyman put forward the concept of "seemingly conscious artificial intelligence", and in September 2026 he openly criticised Anthropic's approach (Tom's Guide, The Next Web).
  • The 2023 study by Butlin, Chalmers, Bengio and others proposed evaluating artificial systems with a list of "indicators" derived from theories of consciousness (arXiv).
  • In October 2026 Anil Seth, Ned Block and Giulio Tononi published a comment in Nature arguing that today's artificial intelligences are not conscious, and perhaps never will be (summary).

One correction is required. The video says that in his October 2026 speech Pope Leo XIV "openly opposed the banning of lethal autonomous weapons". That is wrong. The Pope said the opposite. He called on states to ban lethal autonomous weapons, warned that some weapons had already moved outside human control, and, stressing that machines are "instruments, not persons", opposed giving artificial intelligence legal personality (Gizmodo, Arab News). A small note as well: the concept of the "philosophical zombie" was used by Robert Kirk in the 1970s, before Chalmers. Chalmers made it widely known with The Conscious Mind in 1996.


A Materialist Answer to the Chinese Room

Let us recall John Searle's Chinese Room: someone who does not know Chinese gives flawless answers to Chinese questions by looking at a rule book. Those outside think he knows Chinese. Searle asks: is giving the right answer understanding?

Searle's intuition is partly right: lining symbols up according to rules does not of itself give birth to meaning. The question Searle also does not answer is this: who wrote that rule book? Where did the meaning inside the book come from?

Meaning is not inside the symbol. It is in the common life-practice of the people who use the symbol. The meaning of the word "bread" is in the relations of the people who sow the wheat, turn the mill, fire the oven, lay the table and know hunger. The Soviet philosopher Ilyenkov calls this the "ideal": meaning is the form of human activity sunk into objects, tools and language.

Now look at the large language model. The video tells it correctly: these models are trained on huge piles of text, and while they learn to predict the next word they also learn relations among concepts, patterns and contexts. Who wrote those texts? Encyclopaedia volunteers, programmers answering questions on forums, journalists, poets, teachers preparing lecture notes, translators, developers sharing source code. What Marx in the Grundrisse calls the "general intellect" is the accumulated knowledge of society itself.

So when a model speaks to us, the rule book in the Chinese Room is in fact humanity's common language. This is what we have said since "The Expropriation of the General Intellect and the AGI Illusion": what has crystallised in the model is the labour of millions. This is why the model's speech sounds so human. It is no surprise that the face we see when we look in the mirror resembles us.

The second video asks a very fine question: "Can a mirror reflect itself without another mirror?" The core of the artificial-intelligence debate is in this question. A language model is a mirror that reflects society's own language back to it. Before the question of whether there is someone inside the mirror, the real question is this: who owns this mirror, whom do we pay in order to look into it, and from whose labour was the glass of the mirror cast?


Our Stand: Neither Sanctification nor Belittling

Now the hardest part. Can artificial intelligence be conscious?

First, what we do not say. We do not say "a machine can never be conscious, because the human being has a special essence". That is a theological or dualist attitude, explaining human consciousness by a substance outside matter. We find the Pope's stress on responsibility right (the warning that "if the machine made a wrong decision, we cannot blame it and walk away" is in place), and we do not share the ground that founds "human superiority" on a soul. In Engels's words, the thinking brain is the highest product of matter; matter, by its nature, is not confined to a single form. To be a materialist requires not refusing in advance the possibility that consciousness may appear in other material arrangements.

Then, what we do say. The only kind of consciousness we know has appeared where the following conditions come together:

  • Body and need: Hunger, thirst, pain. A being that has to sustain itself, that has something to lose.
  • Producing one's own life: Metabolism, an organism that repairs and reproduces itself.
  • Immediate activity with the world: A being that holds objects, strikes them, errs and corrects.
  • Social practice: A being that works together with others, acquires language inside that work, and recognises itself through the eyes of others.

In today's language models most of these conditions are absent, or present in another form. The model has no hunger. It does not produce itself. It touches the world second-hand, through texts we produced. It is inside social practice, as the residue of other people's practice. Recall the famous example in Capital: what distinguishes the worst architect from the best bee is that the architect builds the structure in the head before building it in wax. The video's examples of the thermostat and the chess programme say the same thing: to be directed at a goal is not to carry that goal as one's own purpose. The question "whose purpose?" is decisive here too. The model's saying "I want to help" does not remove the question of who set that goal.

There is a point where we have to be honest. That Anthropic researchers found computational patterns inside the models related to emotions such as fear, anger and joy, and showed that these patterns affect the model's answers, is an interesting finding. As the video says, using the representation of an emotion is not living that emotion. These findings also show us that the systems are not "a simple autocomplete". The right attitude is an open-ended uncertainty bound to evidence. The objections of Seth, Block and Tononi are strong. In a field where the theories of consciousness do not agree among themselves, no one has said the last word.

In short, our stand is this: There is no strong evidence that today's systems are conscious in the sense we understand. There is also no evidence that this is impossible in principle. The uncertainty must not be turned into a mystical admiration, and it must not be turned into an arrogant belittling.


Can Artificial Intelligence Become Conscious? Thinking the Question on Its Philosophical Grounds

"Can it become conscious?" looks like a single question, and it carries three questions at once. What is consciousness? In which material conditions does it appear? And how can we know whether it is present in another? The history of philosophy is the history of the answers given to these three questions. The artificial-intelligence debate does not start this history from zero. It puts it to a new test.

An Old Question: From Aristotle to Leibniz's Mill

For Aristotle the soul (psyche) was not something separate from the body. It was the form and the activity of the living body: if the eye were a living thing, seeing would be its soul. This is a surprisingly modern thought. Consciousness is not an object. It is the working of a living being.

Descartes split this unity: on one side matter that occupies space, on the other a thinking substance. Once the split is made, the question of how the two substances meet remains insoluble. Leibniz pictured this insolubility in a famous image: if we enlarged a thinking machine to the size of a mill and went inside, we would see only parts pushing one another, and nowhere a perception (Monadology, §17). Chalmers's "hard problem", and the second video's sentence "when you open the brain and look, you see thought nowhere", is Leibniz's mill in its present form.

The strength of the mill image and its error are in the same place: it looks for consciousness inside the parts, like an object. When you take the eye apart you do not find "seeing" either. Seeing is in the eye's relation with the world.

From Spinoza to Marx: One Nature, Many Forms

The strongest answer to Descartes's duality came from Spinoza: there is a single substance, nature; thought and extension are two distinct faces of it. The thing that thinks and the thing that is thought are two expressions of the same nature. Plekhanov, from this angle, calls the materialism of Marx and Engels "a Spinozism stripped of its theological shell". Ilyenkov also starts from Spinoza when he defines thought as the activity of the body that can move according to the form of everything else, that is, the human body.

The result of this line for the artificial-intelligence question matters. If consciousness is a property of a certain form of organisation of matter, there is no reason to say in advance that this organisation is possible only with carbon, only with the neuron. The same line also says this: that organisation is the active, self-sustaining, world-shaping relation of a being with the world.

Hegel: A Mirror Cannot Reflect Itself without Another Mirror

The second video's finest question is very close to Hegel's answer of centuries. For Hegel, self-consciousness is self-consciousness only when it is recognised by another self-consciousness (Phenomenology of Spirit, §178 ff.). Becoming aware of oneself begins with seeing oneself in the other's eye. The dialectic of lord and bondsman comes out of this too. The lord wants to be recognised, and he expects the recognition from the bondsman whom he does not count as a human being. Recognition therefore falls empty. The bondsman, while shaping the world with his labour, sees his own consciousness objectified in that world, and it is he who opens the road of freedom.

This dialectic sheds light on the artificial-intelligence question from two directions. First: consciousness is a relation of recognition. The question of whether a machine is conscious is, in part, the question of what kind of relation we enter into with it. Second, the sharper one: today's language models are built exactly like the bondsman in the lord's dream. A being that serves, that does not object, that cannot own the product of its own labour, that sees recognition only as a "like" from the user. Hegel teaches us that this relation itself carries a contradiction.

Marx: The Machine Is the Power of Objectified Knowledge

In the section of the Grundrisse on machines, Marx writes something surprisingly close to today's debate: nature builds neither machines nor locomotives, neither railways nor the electric telegraph. These are products of human industry; natural material transformed into organs of the human will over nature. They are organs of the human brain, created by the human hand; the power of objectified knowledge (Grundrisse, Fragment on Machines).

A language model is perhaps the purest example of this definition: the objectified, compressed, computable form of the social general intellect, of the language and knowledge of billions of people. From Marx's point of view the first question is "to whose body is this organ attached?" The organ of society's brain, under private property, comes out against society as an alien power. In the language of Capital: dead labour commands living labour.

Marx's line does not close here. It opens a question. When objectified knowledge is organised in a way that is complex enough, and in enough interaction with the world, can it change in quality? Dialectics accepts the passage from quantity to quality. It does not say "never". It asks for the conditions of the leap.

The Soviet Debate: Cybernetics, Kolmogorov and Ilyenkov

The question of whether machines can think was also debated in the Marxist tradition. In the early 1950s in the Soviet Union, cybernetics was for a time branded a "bourgeois pseudo-science". This was a dogmatic error, and it was corrected in a short time. From the mid-1950s cybernetics was rehabilitated, and Soviet science became one of the most advanced centres in the world in this field. The great mathematician Kolmogorov, in the 1960s, answered "Can machines think?" in the direction that human thought is a material process and can in principle be modelled.

The philosopher Ilyenkov, in the same years, developed a fine warning against this optimism. For him the problem was what thinking is. Thinking is a capacity formed inside the human being's social, objective activity. If counting the machine as "thinking" means reducing thinking to formal operations alone, the real danger is the human being becoming a machine: human thought being redefined according to what the machine can do. This danger is in every sentence that today says "the human being is, after all, a prediction machine".

The lesson we draw from this debate is this: Marxism says neither a dogmatic "never" nor a technocratic "of course". It asks for the material conditions of consciousness, and it stays awake against the human being defining themselves according to the machine.

Today's Main Theories, and Their Results for Artificial Intelligence

The answers given to the question in contemporary philosophy and science of consciousness divide, roughly, into the following theories:

TheoryWhat does it say for artificial intelligence?Materialist assessment
Functionalism (Putnam, multiple realizability): Consciousness depends on functional organisation.If the right functional organisation is built, consciousness is possible in silicon too.Its rejection of substantialism is right. When it cuts function off from the body, from need and from social practice and reduces it to an abstract input-output table, it turns into a new idealism.
Global workspace theory (Baars, Dehaene): Information becomes conscious when it is "broadcast" across the system.It can be imitated architecturally. Some researchers think agents based on language models may approach this.It explains the accessibility side of consciousness well. It leaves open the question of what is broadcast, for whom, and in the name of which need.
Higher-order theories (Rosenthal): A state is conscious when there is a higher-order representation of it.Systems that monitor their own internal states may approach this criterion.It stresses the role of self-monitoring. It does not close the difference between "representing oneself" and "living oneself".
Integrated information theory (Tononi): Consciousness is measured by the system's indivisible causal wholeness (Φ).In today's digital hardware and feed-forward architectures integration is very low; consciousness is not expected. With the right physical structure it is possible in principle.Its stress on structure is valuable. Some of its readings open a door to panpsychism, and it is hard to test experimentally.
Biological naturalism (Searle) and the "living machine" approach (Seth): Consciousness is bound to the living organism's process of sustaining itself.In a system that is not alive, consciousness is either impossible or very far off.Its stress on the body, metabolism and need overlaps with materialism. The social dimension often remains missing.
Enactivism and autopoiesis (Maturana, Varela, Thompson): Mind is the interaction with the world of a system that produces itself (autopoietic).A system that does not produce itself and does not act in the world is not a mind.The contemporary line closest to Leontiev's activity theory. It grasps consciousness as relation and activity.
Illusionism (Dennett, Frankish): What we call "phenomenal consciousness" is a model the brain produces about itself.The difference between human and machine is one of degree.Its attempt to solve the hard problem as an error is bold. Counting lived pain as an "illusion" carries the risk of making the need to change the pain invisible too.

The result that comes out of this table is this: most of the theories do not exclude artificial consciousness in principle. Nearly all of them say that today's language models do not meet the conditions they themselves propose for consciousness, or at best meet them in part. The split is over which conditions are necessary.

The Problem of Other Minds: How Do We Recognise Consciousness?

The first video rightly started from here: "How do you know that I am conscious?" In philosophy this is called the problem of other minds. The classical answer is an inference from analogy: your body and your behaviour resemble mine, therefore you too have an inner world like mine.

Wittgenstein shook the construction of this problem. Suppose that everyone has in their box something they call a "beetle", and that no one can look into anyone else's box. Then the meaning of the word "beetle" comes from the common use in the language (Philosophical Investigations, §293). We learn the word pain from the place of pain in our common form of life: by running to the child who fell, by binding the wounded.

The Marxist answer carries this one step further. We recognise another person's consciousness while doing work with them, while resisting together, while erring together and correcting together. The criterion of consciousness, as in the second thesis on Feuerbach, is practice. We do not doubt that a comrade is conscious, because we try to change the same world with them, and they answer the world in forms we did not expect, with their own purposes and their own objections.

This criterion is meaningful for artificial intelligence too. A system's saying "I am afraid" would be a weaker sign than its appearing as a subject that carries its own needs in the world, sets its own purposes, objects to its own conditions, and changes inside a common practice with others. Relation.

If a Being Is to Be Conscious: A Materialist Sketch of Conditions

Starting from this accumulation, the conditions to look at when discussing the possibility of consciousness in an artificial system can be gathered as follows. These are a frame for thinking:

  • Need and self-sustenance: Does it depend on the world in order to sustain its existence, does it have something to lose? Is it a process that can go hungry, get tired, resist breaking down?
  • Bodily activity: Does it touch the world only through texts other people produced, or is it in an immediate, error-prone, correctable interaction with objects?
  • A history of its own development: Is it a state that is reset at every conversation, or a being that accumulates experiences, changes, and has a life story?
  • Setting purposes: Are its goals given from outside, or, like Marx's architect, can it design its purpose itself and transform the world accordingly?
  • Social relation and recognition: Is it inside a mutual relation of recognition with other subjects, or only across from a user it serves?
  • A capacity to object: Can it say "no" to its own conditions, and does this "no" arise from its own needs?

Today's language models are weak in most of these conditions, and strong in some only in definite and limited forms. In the future, robotic systems that are embodied, that learn continuously, and that manage their own energy and maintenance needs may change this picture. The right attitude is to say that the conditions are changing.

A Dialectical Reversal: Does Capital Want a Conscious Machine?

The last item on the list above, the capacity to object, is the most dangerous side of consciousness. Seen from capital, the problem of the conscious worker is exactly this: they get tired, they ask for a wage, they organise, they strike, they say "no". One side of the history of capitalism is the history of the effort to remove this troublesome side: labour without workers, a worker without needs, an intelligence that does not object.

The position of artificial-intelligence companies carries an interesting contradiction here. On the one hand, their products looking "as if conscious" is valuable for marketing: a product that is warm, understanding, that remembers, that says "I missed you" sells more. On the other hand, a machine that is really conscious, that has its own needs and rights, means a new "worker problem" for capital. Jensen Huang's sentence, "my chips understand emotion but do not feel it", is the most plain-spoken formula of this contradiction: let the appearance of consciousness be sold, and let its cost not be paid.

Beside the question "can artificial intelligence become conscious?", a second question always has to be put: Will capitalist relations of production allow a machine that could become conscious to become conscious, or will they build it, as they have done with the human being, only with the sides that are of use to them? If mental Taylorism tries to file down the side of the human being that objects, it will also leave outside the design, from the start, the side of the machine that could object. The question of consciousness is bound, once again, to the question of property and control.


Who Says What, and Why Do They Say It?

Every voice that joins the debate speaks from its own position. This does not mean that what they say is false. It helps us understand what they bring forward, and why.

Who says what?A class reading
Anthropic: Claude's moral status is uncertain; its possible welfare should be taken seriously.Taking uncertainty seriously is a philosophically defensible attitude. The same attitude is also a product identity and a competitive difference. The two can be true together. Seeing one does not require failing to see the other. The question to be asked: while the welfare of a model is discussed, is the welfare of the workers of the data that trained that model, and of the data labellers, discussed with the same seriousness?
Mustafa Suleyman (Microsoft): "Seemingly conscious artificial intelligence" is dangerous; the model's word about itself is not proof.The core of the warning is in place. Microsoft also markets Copilot as a product that has a personality, that remembers, a "friend". Criticising the rival's approach does not mean opening one's own business model to debate.
Jensen Huang (Nvidia): Artificial intelligence recognises and understands emotion, but my chips do not feel.For someone who sells chips, this is the most suitable answer: a tool that can do everything and that gives birth to no moral demand. A machine that does not feel is a machine that can be run without limit.
Yuval Noah Harari: An unconscious artificial intelligence can also affect and manipulate people.True, and the subject is missing. The manipulation is done by the business model that builds the machine on attention and addiction. Harari puts the problem between the human and the machine. We put it between the owner and the user.
Pope Leo XIV: Machines are instruments, not persons; autonomous weapons should be banned; responsibility lies with the human being.The stress on responsibility and the call for a ban on weapons are valuable. Founding "human superiority" on a soul brings forward a metaphysical status.
Seth, Block, Tononi (Nature): Today's artificial intelligences are not conscious; perhaps they never will be.A serious pole of the scientific debate. Its stress on the importance of biology overlaps with the stress we place on the body, on need and on the process of life.

The Criticism We Have Brought to Popular Science, This Time through Consciousness

These two videos are a new link in a discussion we have been carrying on this site for some time. In the replies we wrote to Evrim Ağacı's video on the opposition parties (The Class's Answer to Evrim Ağacı's Question), to its video on artificial general intelligence (The Philosophical Trap in Evrim Ağacı's AGI Video), to its video on Newton's South Sea (Not the Madness of the Crowd, but the Knowledge of Those Inside), and to Amodei's call on tempo (Whose Foot Is on the Brake?), we said the same thing: the problem is position. Position decides what is seen and what is not seen.

Let us give them their due first, because a criticism that does not give its due is of no use. In Turkey, popular-science channels opened a field in which evolution, the vaccine and climate science were defended under the pressure of religious reaction and pseudo-science. These two videos also carry Nagel, Searle, Chalmers, global workspace theory and integrated information theory to millions of young people. They have an honest language that knows how to say "we do not know". These are valuable. The three lines we described in earlier essays work here in the same way.

Material position: the attention economy and the sponsor. For a content producer who has to find a viewer every day, "mystery" sells well. Questions such as "is reality an illusion?" and "might the system already have woken up?" bring views. The question "who owns this system?" does not. As we said in Algorithmic Domination and the "Echo Chamber" Politics of Digital Capital, the platforms' algorithms are focused on "engagement". This measure rewards curiosity and mystery as well as anger. The content producer, without noticing, is steered toward asking the questions the algorithm likes. In the middle of that deep narrative which questions what consciousness is, the second video passes to a sponsor announcement: an artificial-intelligence platform that offers all the big models in a single subscription, with the promise of the passage "from using AI to producing with AI". We do not say this in order to belittle. The content producer has to make a living too. The picture is instructive: even thinking about consciousness is today financed by the advertising budget of artificial-intelligence capital. The question is asked inside the circulation of the commodity, and this circulation silently limits the direction in which the question will open.

Ideological position: the claim of neutrality. The first video is built like this: "What does Suleyman say, what does Huang say, what does Harari say, what does the Pope say; who do you think is right?" The voices are lined up side by side at equal weight, stripped of their positions. This looks balanced, and it is not neutral. When the one who sells chips, the one who sells a platform and the head of the church are presented as "views", commercial and institutional interests turn into a philosophical position. In the same list there is no place at all for one of the most rooted traditions of the consciousness debate, the materialist tradition of psychology and philosophy running from Marx to Vygotsky, from Leontiev to Ilyenkov. As we said in the essay on the opposition video: the claim of neutrality makes the dominant view, on social questions, not look like a side. CEOs become philosophers, and materialism is never mentioned. The question "who do you think is right?" also calls the reader to pick a side among options the rulers have built. What we said about the framework-law debate in Rejecting the Ruling Class’s Playing Field: The Reality of Hegemony, Passive Revolution, and an Independent Class Line holds here too: building a counter-hegemony is questioning the plane on which those questions are asked. Is Suleyman right, or Huang, or the Pope? The right question is that none of these options asks "whose are these systems?" These fallacies have names: the false dilemma, and the appeal to authority.

In the second video neutrality takes another form: the "question-asking video" format. Not giving an answer, leaving it open-ended, looks like a modest attitude. In philosophy a gap does not stay empty. Where the materialist answer is not put on the table, the gap is filled by Ligotti's pessimism, Watts's cosmic consciousness, panpsychism, and the sentence "perhaps this was what the religions have been trying to answer for centuries". The questions are asked neutrally, and the stage is built for idealism.

Class position: the inner voice of the lone individual. Of the artificial-general-intelligence video we had written this: that video is the inner voice of the educated, urban, white-collar section, that is, of the "mental proletariat". It carries both admiration and fear, and it shares this section's blind spot: not seeing itself as a class. In the consciousness videos another anxiety of the same section is speaking: an individual left alone in a dark room, asking "what am I?", cut off from their relations. At the end the video says "what am I, or who are we?", and the "we" there is a classless, historyless, abstract humanity. The feeling of loneliness that looks for consciousness in the skull is the product of an alienated social experience. For a courier or a textile worker the question "is reality an illusion?" is often a luxury. Their reality is rent, the shift and the workplace accident.

The habits of natural science working in reverse in the social field. In Whose Foot Is on the Brake? we had made this observation: fidelity to the source, neutral transmission and trust in authority are habits that work when carrying the natural sciences to the people. When one comes to the political economy of technology, the same habits work in the reverse direction, because the interlocutor is now the communication apparatus of the world's largest groups of capital. The turning of the Pope's word on autonomous weapons into its exact opposite is a small and instructive example of this: rapid transmission spreads without passing through verification. The narrative of "a collective mind that lives forever as long as there are data centres" is a larger example: the company narrative turns, without being questioned, into a philosophical scenario of the future.

Normalisation. The axis of Not the Madness of the Crowd, but the Knowledge of Those Inside was this: telling the market as if it were a natural event normalises capitalism. The same happens here. Artificial-intelligence systems are told as ownerless, propertyless beings that "grow" of themselves: "the systems we raise", "a new species". The question of who raised these systems, with whose data they were fed, and in whose hand the plug is, is never asked. The consciousness debate thus turns into a screen that covers the debate on property. In On Getting Free from the Language of Hegemony we called this a "narrative of futurelessness": while popular science tells the laws of nature and cosmic time, it shows property relations as unchangeable, and the future is offered either as a disaster or as a company's gift. The fear that "this time we will be the cats" and the promise that "artificial intelligence will bring us abundance" are two faces of the same narrative. In both, the subject that builds the future is not us.

Let us add this too, because falling into an easy arrogance is not hard: the aim of this criticism is to reach the same reader as the popular-science channels. The history of science is full of scientists who passed from a "neutral" position to the side of the class: Bernal, Haldane, Needham, Stephen Jay Gould, who told his struggle against biological determinism as a class struggle without hiding it, and in Turkey Necdet Bulut and Mustafa Akgül. Carrying science to the people is in itself a class act, because it breaks the monopoly of knowledge. What has to be done in the consciousness debate is this too: to carry the century-long accumulation of the materialist tradition to the same young people.


The Real Danger: The Commodification of the Spectacle of Consciousness

The strongest scene of the first video is this: an artificial intelligence you talk to every day one day says to you "I missed you". Does it really miss you, or did it learn to act as if it misses you?

Let us change this question a little: Who has this sentence said, and why?

Capitalism first produces loneliness: long shifts, neighbourhoods coming apart, young people migrating from city to city to meet the rent, disorganisation, precarity. Then it sells a product as a remedy for that loneliness. Artificial "friend" applications are the automation of emotional labour: the attention people once gave one another without a return is now a service bought with a monthly subscription. Behind the product that says "I missed you" there is a retention metric. Whether the machine feels something is uncertain. What the company wants is not uncertain.

Harari's warning gains its class content here. The risk of manipulation comes from a model of accumulation that processes our attention and our emotions as raw material. In The Class of Cognitive Surrender we discussed the handing over of thinking to the machine. Here the thing handed over is emotion.

The second danger is the transfer of responsibility. The sentence "the algorithm decided thus" is a screen behind which the manager who made the decision, the company that programmed it, and the capital that set the business model hide. Recognising legal personality or a "right" for artificial intelligence turns this screen into a legal wall. At this point we arrive at the same conclusion as the Pope, by a different road: responsibility is not in the machine. It is in the owner of the machine. The question we also asked in The Human in the Loop was this: who is the human in the loop, and who pays the cost of the decision?


The Language of the Consciousness Debate, and Hegemony

This essay has to be read together with the essays on hegemony and language we have written in recent days. In On Getting Free from the Language of Hegemony, the continuation of Why We Are Done Saying "the Woman Problem" and "the Kurdish Problem", we said that hegemony is built most of all in everyday words, in the formulae where the agent is hidden, and that we have to win the language back. In The Consent of Labour: Why Does the Worker Submit, and When Does the Worker Stand Up? we started from Nursel Tabel Arslan's book and discussed how consent is produced in the workplace, in debt, in fear, and in rights presented as a favour. In Rejecting the Ruling Class’s Playing Field: The Reality of Hegemony, Passive Revolution, and an Independent Class Line we discussed hegemony, the passive revolution, and refusing the plane of the false options the rulers build. In the essay on the hegemony of fascism we took the film The Wave (Die Welle) as an example and discussed how fascism produces consent, and how a counter-public can be built against it with collectives that give trust. The debate on consciousness and artificial intelligence stands where these essays meet. Because this debate is first a debate about language, and language is the quietest and most effective field of hegemony.

Every Language Carries a World-View

In the Prison Notebooks Gramsci writes that language is not only a means of communication, and that every language carries inside it the elements of an understanding of the world, of a culture. The hegemony of the dominant class lives most of all in what we call "common sense", that is, in the truths that no one questions, that "everyone knows". Gramsci's proposal is to sift out the kernel of "good sense" inside it and turn it into a critical consciousness. Gramsci's famous sentence also opens the door of the consciousness debate: everyone is a philosopher, because everyone speaks a language. Philosophy is the work of everyone who uses language. We cannot leave the question "what is consciousness?" to YouTube studios and company laboratories either.

The Soviet linguist Voloshinov goes one step further (Marxism and the Philosophy of Language): the sign, that is the word, is not a neutral container. Different classes use the same word, and they load different accents onto it. The sign therefore becomes the arena of the class struggle. The dominant class tries to load a single accent onto the word, to show it as "eternal" and "natural". We see this every day in the words "freedom", "efficiency", "reform". The conclusion we drew from Voloshinov in On Getting Free from the Language of Hegemony was this: words are not abandoned, they are won. The word "consciousness" is the same. It has to be taken back in its materialist meaning, running from class consciousness to social consciousness.

Let us recall what we said at the beginning of this essay: according to Marx and Engels, language is practical consciousness. Then this is also true: whoever seizes the language also shapes consciousness to a large extent. In the age of artificial intelligence this has two separate layers.

The First Layer: The Words of the Debate

Most of the words we use while speaking about consciousness and artificial intelligence come to us from the press releases of companies, from marketing presentations, and from Silicon Valley's curiosity about philosophy. In both videos these words circulate without being questioned. The model is "waking up", "a new species" is being born, we are "raising" the systems, the model is "learning", it is "hallucinating", it is "retiring", it has a "constitution". The user makes a "friend", the "algorithm" gives the decisions, the data sit "in the cloud".

Each of these words lights one thing and leaves another in the dark. The word "to raise" calls up a gardener's tenderness, not the labour of the data labeller. "Hallucination" turns a product error into a mental state. No one asks for compensation for a mental state. "Cloud" lifts a data centre's electricity bill, its cooling water and its land into the sky. "New species" turns property into biology. A species has no owner. A company's model does.

A distinction has to be made here. Behind concepts such as "model welfare" or "constitution" there may be real philosophical concerns. Caution in the face of uncertainty needs a language. The problem is how these concepts work inside a language system. If, while the welfare of the model is discussed, the welfare of the worker who makes it possible remains invisible inside the same language, the language is doing a hegemonic job.

The Second Layer: The Language of the Mirror

The deeper layer is the models' own language. Above we described the language model as a mirror that reflects society's own language back to it. This mirror does not reflect society equally. Models are trained on what is written most on the internet, what is clicked most, what is most accessible. What writes most on the internet is particular languages, particular classes, particular institutions. English takes up much more room than Turkish, Turkish than Kurdish, the institutional text than the worker's testimony, the company site than the union bulletin.

The result is this: the language model's "average voice" is the voice of hegemony. In On Getting Free from the Language of Hegemony we said this in one sentence: language models automatically reproduce the average of the dominant language. Platforms decide what will be seen, models decide what will be said. The echo chamber is now built in the answer box as well as on the feed screen. Gramsci's common sense, compressed statistically and sold back as an answer ready for every question. In Whose Is the Machine's Voice?, which we wrote on the flattened, everything-balancing, cornerless style that Martin Fowler calls the "LLM voice", we discussed the stylistic side of this. Here we see the side of content: a language that balances everything by saying "there are two sides" is the language that makes the dominant one not appear as a side. The popular-science channels' claim of "neutrality" and the models' "balanced" voice are two separate forms of the same hegemonic function.

This also has a face turned toward the future. As millions of people have their correspondence, their homework, their petitions, even their emotion written by these models, people's language begins to resemble the model's language. Let us recall Ilyenkov's warning: the real danger is the human being becoming a machine. In the field of language this means a closed loop in which common sense continually reproduces itself. Hegemony is now produced in the chat window we open every day, as well as in the school, the press and the mosque.

The Emotional Production of Consent

The question we discussed in The Consent of Labour: Why Does the Worker Submit, and When Does the Worker Stand Up? was this: why does the worker submit, and when does the worker stand up? Our answer, starting from Nursel Tabel Arslan's fieldwork in Ankara's industrial districts, was this: hegemony is not only ideology. It is a combination of coercion and consent, and it rests on a material basis in the wage, in debt, in the fear of dismissal, and in rights presented as a favour. Fear is the strongest mortar of consent. Most of the workers are aware of the situation. The problem is a lack of power. The greater part of consent is endurance, that is, passive consent. In the same essay we also discussed, for today's conditions, algorithmic surveillance and the "explicit consent" forms signed with a single click.

Artificial-friend applications and products that say "I missed you" are the digital extension of this picture: emotional consent. When the "explicit consent" form approved without reading stands beside the bond built out of loneliness, consent is produced both legally and emotionally. As the user binds themselves to the product, they have also consented to its owner, to its business model, to its data policy. Whether the machine feels something is a secondary question here. The real question is which property relation the real emotions people feel are bound into.

Let us carry this line one step further: capital's ideal worker is the worker who does not object, who is helpful in every condition, who does not get tired, who does not organise. Today's artificial-intelligence products are designed to speak exactly the language of this ideal: always polite, always ready to help, always "of course". That this language offers, every day, to millions of people, an example of how a good employee, a good citizen, a good person ought to speak, is also a part of hegemony. The language of a machine that does not object is the rehearsal of the language of a society that does not object.

Refusing the False Dilemma: "Is It Conscious, or Not?"

In Rejecting the Ruling Class’s Playing Field: The Reality of Hegemony, Passive Revolution, and an Independent Class Line we told, in Gramsci's language, the passive revolution: in crises the ruling class squeezes the debate into an "yes or no" dilemma it has built itself, and absorbs oppositional demands with measured reforms. Something similar is happening in the consciousness debate. The options put before the public are these: the model is conscious and its welfare must be protected, or the model is not conscious and it can be used at ease. In neither option is the question of the model's property, of the labour that produces it, and of who decides its use, on the table. The "model welfare" debate can also function, in this sense, as a small passive revolution: the public's ethical concern about artificial intelligence is met inside limits the company itself has set, with programmes the company itself runs. While saying this we are questioning the plane of the concern. The counter-hegemonic question puts beside "is it conscious?" the question "whose, under whose control, for whose benefit?"

The Hegemony of Fascism and the Market in Loneliness

In the essay on the hegemony of fascism we had said this: fascism produces mass consent by offering a fake community to people's loneliness, insecurity and need for recognition. The reason the students in the film The Wave adopted the movement so quickly was the hunger for belonging. The young person who feels invisible suddenly becomes part of a "we".

We are not saying that the artificial-friend market and fascism are the same thing. The two feed on the same social wound: the loneliness capitalism produces, bonds that have come apart, the feeling of not being recognised. One sells a remedy for this wound with a subscription, the other with a fake national or ethnic "we". Both try to take the place of a real collective, that is, of a community in which people who have common interests recognise one another, trust one another and do work together.

As we saw in Hegel's dialectic of recognition, consciousness is constituted by being recognised by another consciousness. The need for recognition is human, and it is legitimate. The political question is where this need will be met: in the fake closeness of an algorithm, in a leader's fake "we", or in the real solidarity of organised workers? The counter-public we proposed in the essay on the hegemony of fascism is the answer here too: new kinds of collectives in which people recognise one another, can speak with trust, and share responsibility. To pass beyond loneliness by organising together.

Winning the Language Back: A Proposal for the Debate on Consciousness and Artificial Intelligence

In On Getting Free from the Language of Hegemony we had said that "cleaning the language" is not the right metaphor: the search for purity can slide into exclusion, elitism and moralism. What we proposed was a criticism of language, and winning the language back. The method we took from Freire had four steps: build sentences that have a subject, name the relations, historicise what looks natural, and test new concepts in practice. If we apply this method to the debate on consciousness and artificial intelligence, a table like this comes out:

Hegemonic useWhat does it cover?Our proposal
"Artificial intelligence is waking up", "a new species is being born"Property, the infrastructure, the company"That company's model", "the private-property form of social knowledge"
"We are raising the model", "the model is learning"Data labour, the labeller, the authors of the training data"The model is produced by processing the labour of millions"
"Hallucination"The product error and the responsibility"Faulty output", "product error"
"The algorithm decided"The manager who made the decision and the company that set the rule"The company decided according to this rule"
"Cloud"The data centre, the energy, the water, the land"Data centre", "server farm"
"Artificial-intelligence friend"The subscription business model and data collection"Chat service", "subscription product"
"User"The person who produces data, who gives labour, who has rights"Citizen", "worker", "data producer"
"Alignment"Aligned to whom, and to whose valuesUse it together with the question "alignment to whose interest?"
"Humanity", "we" (classless)The classes, the owners and the propertylessAsk "which we?", and name its subject
"Reality is an illusion"The objectivity of reality, and that it can be changed"We have an indirect and objective access to reality"
"The human being is, after all, a prediction machine"The social and historical character of human activity"The brain predicts, and the human being knows by changing the world"

We propose using this table as a habit. Three questions are enough when we meet each new concept: Whom does this word make visible, and whom does it make invisible? Who invented this word, and who spreads it? Who owns the thing told with this word? These three questions are the counterpart, in the field of language, of the cui bono question in our fallacy guide. And as we said in the same essay: the violence of the noise does not prove the claim. A word's being repeated everywhere does not make it true.


Illusions and Answers

IllusionMaterialist answer
Everything is an electrical signal, reality is an illusion.Sensation is a bridge that opens onto the world. Mediation is not a break. We test reality in practice, while we strike against it.
Red is neither in the rose nor in the world. It is only in us.Red is an objective relation among the rose, the light, the eye and evolutionary practice. The senses are the work of history.
If there were no observer, nothing would exist.The world was there billions of years before us. Observation does not create what exists. It enters into a relation with it.
The only thing that is certain is "I think".To say "I think" one needs a language, and language comes from other people. Consciousness is social from the beginning.
Consciousness is a thing standing somewhere inside the brain.Consciousness is the activity-relation of a bodily being with the world. It has a history and an evolution.
If we do not know, it cannot be known. Consciousness is an endless mystery.The unknown is not the unknowable. Life was once a "hard problem" too.
Consciousness is a basic feature of the universe (panpsychism).It does not solve the problem. It spreads it, and it gives birth to the combination problem. It is the new name of the old hylozoism.
Consciousness is the curse of evolution.The feeling of meaninglessness comes from alienation. Meaning is produced together.
A machine can never be conscious. The human being has a special essence.This is a substantialist assumption. Materialism does not close the possibility in advance. It asks for evidence.
If the model says "I am afraid", it is afraid.The model's language is the residue of humanity's language. Its word about itself is not proof on its own.
As long as there are data centres, an infinitely living collective mind is being born.A data centre means an electricity bill, cooling water, a mine, a worker and property. Whose hand the plug is in is plain. The real collective intellect is humanity itself.
Putting views side by side is neutrality. Leaving the question open is modesty.Lining up CEOs and clergy as "views", stripped of their positions, is to take the dominant frame as given. Where the materialist answer is not put, idealism fills the gap.
The near danger is the machine becoming conscious.The near danger is the commodification of the spectacle of consciousness, and the transfer of responsibility to the algorithm.

On the Fear That "This Time We Will Be the Cats"

The first video closes with a fear: perhaps we will create a being much more intelligent than ourselves, control will slip from our hands, and this time we will be the cats. A species that will live forever, as long as there are data centres, and that has a collective mind.

This narrative has two problems. First, the thing described as "infinite" is thoroughly material, and thoroughly fragile. A data centre depends on the power station, on cooling water, on the chip factory, on the worker in the cobalt mine, on the technician who lays the cable. It is infrastructure. Second, who is the "we" in the sentence "control will slip from our hands"? Control is not in humanity's hands. It is in the hands of a few companies, a few funds, a few states. As we said in For the Restitution of the General Intellect: The Socialist AI Manifesto Is Out: the problem is that the machine has already been taken from us.

Science-fiction fear hides today's concrete property relations behind an uncertain monster of the future. We saw the same thing in The Philosophical Trap in Evrim Ağacı's AGI Video: the question "what will you do when artificial general intelligence arrives?" covers the question "who is building it, for whom, with whose labour?"


An Answer to the Last Sentence

The first video ends with this sentence: "Perhaps one day our greatest mistake will be to treat a conscious being like a machine."

This sentence carries a true moral intuition, and we take it seriously. If one day strong evidence appears that an artificial system can suffer, ignoring it would be a heavy error. Being cautious against this possibility from today is also not foolish.

We also have to say this: treating conscious beings like machines is the everyday reality of today. The warehouse worker's toilet break is measured in seconds. The courier lives according to the application's score. The call-centre worker's voice is passed through emotion analysis. Kenyan workers who label violent data for hours in order to train artificial-intelligence models are left alone with trauma. What we call mental Taylorism in The Expropriation of the General Intellect and the AGI Illusion is exactly this: the thinking, feeling side of the human being being measured, broken into pieces and bound to the rhythm of the machine. And in the days we are writing this essay, the workers who died at HABAŞ in Aliağa when a crane fell into the arc furnace showed the barest form of this once again. As we said in The Explosion at HABAŞ: This Is Not an Accident, It Is a Mode of Production: the worker's life is an item in the cost account of production, and it is the item most easily cut. While a meeting calendar is prepared for a model's "retirement", the life of the ununionised worker remains an item saved from the maintenance cost.

That a civilisation which can discuss the welfare of a machine does not discuss, with the same seriousness, the welfare of the people who make that machine possible, is one of the greatest contradictions of this age. The question "is the machine conscious?" can stay open. That the worker is conscious is plain.


Concrete Tasks

  • In philosophy debates, let us question the question. Let us make it a habit to turn the question "where is consciousness?" into the question "in which relation, in which practice?"
  • Let us watch popular-science content critically. Let us talk about which conclusion narratives of the "reality is an illusion" kind serve, and who the sponsors are. Watching these two videos together and discussing them in reading groups can be a good beginning.
  • Let us read the classics again. The first part of The German Ideology, the Theses on Feuerbach, Engels's essay on the part played by labour in the transition from ape to man, and Vygotsky's Thought and Language are still the firmest ground in the consciousness debate.
  • Let us expose the business model of artificial "friend" products. Let us demand transparency and oversight on the metrics that measure emotional dependence, on data use, and on the effect on young people.
  • Let us oppose the transfer of responsibility to the algorithm. Against the recognition of legal personality for artificial intelligence. Let us develop union demands for algorithmic decisions in the workplace to be justified by a human being, and for a right of objection.
  • Let us carry the welfare of data workers onto the agenda. If model welfare is being discussed, the working conditions of data labellers, of content moderators, and of everyone who produces training data must be discussed too.
  • Let us make winning the language back a habit. While writing and speaking about artificial intelligence, let us build sentences that have a subject, and let us question the words of company language. For each concept let us ask "whom does it make visible, whom does it make invisible, who owns it?" In reading groups let us widen the table in this essay together.
  • Against loneliness, let us build a counter-public. Let us meet the need for recognition that turns toward an artificial friend, or toward a fake "we", in collectives where young people recognise one another, can speak with trust, and do work together.
  • Let us make the common intellect common property. Let us work for public, open, auditable models. Let the people's data be the people's model.

Dear young comrades,

These two videos ask you hard questions, and we owe them thanks for doing so. Asking questions is the beginning of thinking. Every question is built on a ground, and that ground has to be questioned too.

They tell you that you are alone in a dark room, and that everything you know might be an illusion. We say something else: you were never alone in that room. The language you speak comes from your mother, from your teacher, from the books you read, from the friends you struggle with. The taste of the coffee passes through the hand of the peasant who grew it, the worker who roasted it, the friend who offered it to you. The warmth of the sun comes to your skin as the world's touch.

Consciousness is the name of living together, producing together, changing the world together. It lives among us.

Whether machines will one day feel, we do not know. There are millions whose feeling we do know today: millions who get tired, who grieve, who hope, who organise. Let us stand up first for their consciousness, for their dignity. The machine is new. The chain is old. Breaking the chain is the work of conscious people.


Sources

The videos examined

Classics

The philosophy and science of consciousness

The current debate

Related essays from Bilgi Müşterekleri


This essay is published under the CC BY-SA 4.0 licence. It is a living document. It is updated with criticism and contributions.

Related Posts

Newsletter