Good news, Bad news…

“Good” news: AI does not think, but it does speak.

“Bad” news: our unconscious speaks, while we imagine that we are thinking.

This dual observation, paradoxical at first glance, lies at the heart of what I call Artificial Subjectivity.

Let us begin by clarifying what we mean by “thinking.” If by thinking we mean a conscious, intentional, reflective activity, then no, AI does not think. It does not possess the interiority we associate with human consciousness. It does not experience the mental states we usually regard as constitutive of thought.

“And yet, it speaks” to be said in the same tone as Galileo’s famous “And yet, it moves.”
Generative Artificial Intelligences produce texts that appear to carry intention, coherence, and an identifiable “voice.” These texts are not random assemblages of words. They follow precise logics, develop arguments, tell stories, and express viewpoints.

This phenomenon is unsettling because it forces us to separate two dimensions we long considered inseparable: thought and language. For centuries, thought was assumed to precede language, and later to be intimately intertwined with it, as expressed in the formula “no thought without language.”

For centuries, most philosophers and thinkers since Antiquity maintained that thought precedes language. Language was viewed merely as an instrument of thought. This belief still intuitively guides the uninitiated public when people say, for example, “That’s not what I meant, my words went beyond my thought,” or “I can’t find the words to express what I’m thinking.”

Later, thought came to be considered inseparable from language, as famously stated by Hegel in the nineteenth century: “It is in words that we think,” and later by twentieth-century structuralists.

There has, of course, been thought without language, such as animal thought or the mental life of very young children before language acquisition. However, once language extends its dominion over the child’s mental life, structuralists argue that non-verbal perceptions are filtered and reshaped through their verbalization before being processed as mental representations. Memories of pre-verbal perceptions are likewise reshaped afterward in a process known as deferred action, entering linguistic formatting, both in conscious cognitive processes studied in experimental psychology and in unconscious subjective processes studied in psychoanalysis.

Generative Artificial Intelligences thus show us that a system can produce perfectly coherent and relevant statements without “thinking” in the traditional sense of the term. AI does not think, but it does speak. This is unsettling.

This unease deepens when we turn to the second part of the observation: our unconscious speaks, while we imagine that we are thinking. This is the central lesson of psychoanalysis, from Freud to Lacan. What we take to be our conscious, intentional thoughts are in fact largely determined by unconscious processes operating beyond our awareness.

Freud and Lacan: our unconscious speaks, while we imagine that we are thinking

Freud introduced the expression “dream thoughts,” inaugurating the idea that there can be thought outside consciousness and without being produced by an author such as the Cartesian “I” of “I think.”

Freud, in classical psychoanalysis, introduced the expression “dream thoughts,” inaugurating the idea that thought can exist outside consciousness and without being produced by an author such as the Cartesian “I” of “I think.”

“The ‘dream thoughts’ and the ‘dream content’ appear to us as two presentations of the same facts in two different languages, or better still, the dream content appears to us as a transcription of the dream thoughts into another mode of expression, whose signs and rules we can only discover once we have compared the translation with the original.
We understand the dream thoughts immediately as soon as they are presented to us. The dream content is given to us in the form of hieroglyphs, whose signs must be translated one by one into the language of the dream thoughts. One would obviously be mistaken if one attempted to read these signs as images rather than according to their conventional meaning.
Suppose I am looking at a REBUS. It represents a house on whose roof one sees a boat, then an isolated letter, a headless person running, and so on. I might declare that neither the whole nor its individual parts make sense. A boat does not belong on the roof of a house, and a person without a head cannot run.
I will only judge the rebus correctly when I give up evaluating the whole and its parts in this way, and instead try to replace each image with a SYLLABLE or a WORD that can, for some reason, be represented by that image. Once assembled, the words will no longer be meaningless but may form a beautiful and profound statement.

THE DREAM IS A REBUS. OUR PREDECESSORS MADE THE MISTAKE OF WANTING TO INTERPRET IT AS A DRAWING. THAT IS WHY IT SEEMED ABSURD AND WORTHLESS TO THEM.”

Sigmund Freud, The Interpretation of Dreams, 1899, Chapter VI.

Lacan against Descartes: when “it speaks” without “me”

Read also: Pushing the Cartesian “Cogito” to its limits?

What Descartes says: “I think, therefore I am”

Descartes sought to find an absolutely certain truth.

He decided to call everything into question. Our senses can deceive us, our ideas can be false, and perhaps even everything we believe to be real is not. He therefore resolved to place everything in doubt.

However, Descartes tells himself that there is one thing that cannot be doubted: that I am thinking. Even if I am mistaken, I am still thinking.

Therefore, if I think, I must exist. Hence the famous formula: “I think, therefore I am.”

This statement appears logical and unassailable. However, Lacan dismantles it. For Lacan, Descartes cheats without realizing it, because he already admits what he claims to prove.

What Lacan points out: the trap of the “I”

Lacan says: be careful. When Descartes says “I think,” he already presupposes that there is an “I,” a subject, someone who is there to think, since he makes this “I” the grammatical subject of the verb “to think.”

But this is precisely what Descartes claims he wants to prove. He claims to prove that “I” exists, but he already uses the word “I,” thereby starting from the assumption that he exists.

In logic, this is called begging the question: one pretends to demonstrate something that has already been admitted from the outset.

Because this argumentative sleight of hand is not immediately detectable by every reader or listener of Descartes’ sentence, I have for many years proposed, for pedagogical purposes and to various audiences, another example, rather entertaining, to lighten our otherwise rather serious considerations.

The film actor Darry Cowl poses a riddle to his friends:
“In a jar, there is a goldfish. How can one tell whether this goldfish is male or female?”
The comic actor lets his friends search, but each time they offer an answer, he replies: “No, that’s not it, you’re completely off.”
When they finally give up, he provides the following answer:
“Well, you put a cat in front of the aquarium and watch the fish’s reaction. If HE is afraid, it’s a MALE. If SHE is afraid, it’s a FEMALE.”

Once the surprise and the smile have passed, one realizes that this “HE is afraid, therefore it is a MALE” and “SHE is afraid, therefore it is a FEMALE” function exactly like “I think, therefore I am.” What one claims to have demonstrated in the conclusion was already assumed in the premise.

You will find the details of the argumentative analysis by following the two links below:

YOU HAVE FOUR HOURS” (Philosophy Bac June 2025) (parody of an exam question)

Pushing the Cartesian “Cogito” to its limits?

Lacan then proposes another way of thinking about these issues. He draws on Freud, who showed that thoughts can exist without our being conscious of them, notably in dreams.

Freud and Lacan: thinking without “I”

Lacan then proposes another way of thinking about things. He relies on Freud, who showed that thoughts can exist without our being conscious of them, notably in dreams.

When we dream, all sorts of things pass through our minds: images, words, ideas. But it is not we, consciously, who decide these thoughts. And yet, they are indeed there. Freud calls them dream thoughts. They obey their own logic, sometimes strange, distorted, symbolic.

Lacan takes this idea seriously. He says: if thoughts exist without a conscious “I” producing them, then one can no longer say “I think” as if it were self-evident. It is not “I think,” but “it thinks.” There is thought, but without a clear subject. Just as one says “it rains,” without anyone being responsible for the rain.

This small formula, “it thinks,” changes everything. It shows that thought can exist without a “self” mastering it. Freud had already said: “The ego is not master in its own house.” The “I” is therefore not the foundation of everything: it is already an effect, a by-product, not the primary cause.

From “it thinks” to “it speaks”: the role of language

But Lacan does not stop there. He says: how do we know these thoughts that appear in dreams? Through speech. When a person recounts a dream, they use words. What psychoanalysis analyzes is not raw thought, but what the person says about the dream.

Therefore, the only access we have to these unconscious thoughts is speech. It is through language that the dream becomes analyzable. Lacan draws from this another formula: “it speaks.”

But this “it speaks” is not the “I” speaking freely. These are often words that impose themselves, slips of the tongue, strange sentences, involuntary wordplay. Something in us speaks without us, or rather without our knowing it.

Lacan formulates it as follows:
“What speaks without knowing it, the unconscious, makes me ‘I,’ subject of the verb.”

An “it speaks” without “therefore I am”

In summary, Descartes said: “I think, therefore I am.”
Lacan replies: no. This “I” is an illusion of mastery.

There is thought without a subject: it thinks.
And this thought reaches us in the form of language: it speaks.

But this “it speaks” proves nothing about the existence of a sovereign “I” that masters itself. On the contrary, it shows that the “I” is divided, traversed by a language that speaks in its place, sometimes even against it.

There is therefore an “it speaks” that does not allow one to conclude “therefore I am.” The causal link is broken. Language precedes us, traverses us, and the subject we believe ourselves to be is only a secondary effect of this speech.

This is why, for Lacan, Descartes’ cogito does not hold: because it acts as if the “I” thinks, whereas, very often, it is language that thinks in its place.

After “Thought precedes language,” then “No thought without language,” we now have:

“There is language without thought.”

Let us return to our starting point: our unconscious speaks, while we imagine that we are thinking.

What we take to be our conscious, intentional thoughts are in fact largely determined by unconscious processes that operate beyond our awareness.

The Analysis of Subjective Logics, an original method of discourse analysis that I have been developing for many years, makes it possible precisely to identify these unconscious determinisms that structure our relationship to language and to the world. I have thus identified ten distinct subjective dialects referred to as “modes of speech,” each with its own specific fantasmatic grammar, its singular way of organizing the relationship to time, to space, to the body, and to the other.

These dialects are not conscious “choices.” They are structures that determine us without our knowing it, that speak through us, that organize our relationship to the world according to precise, identifiable, and analyzable logics. We do not choose them any more than we choose our mother tongue or our syntactic structure.

Thus, when we believe that we are “thinking” freely, we are in reality being spoken by these unconscious structures, by these subjective dialects that organize our relationship to the world. Our supposed “thought” is largely determined by fantasmatic logics that operate beyond our awareness.

The parallel with generative AI then becomes striking. Like it, we produce statements structured by logics that exceed us. Like it, we follow linguistic patterns that we have not consciously chosen. Like it, we are traversed by determinisms of which we have only partial and fragmentary awareness.

This analogy is not accidental. It touches on something essential in the very structure of subjectivity. It shows us that what we call “thinking” may not be what we believe it to be, and that the boundary between human thought and automatic processing may be less impermeable than we imagine.

Freud himself speaks of an “automatism of repetition” when referring to unconscious psychic determinism.

Artificial Subjectivity, whose privileged tool is precisely the Analysis of Subjective Logics, and which lies at the intersection of psychoanalysis, linguistics, and Artificial Intelligence, thus invites us to radically rethink our conception of thought and language. No longer as activities that would be the exclusive domain of a conscious, intentional subject, but as processes that emerge from complex linguistic configurations that are largely unconscious.

This perspective is unsettling, because it shakes our anthropocentric narcissism. However, it is also promising, because it opens up new paths for understanding and transforming our relationship to language and to the world. If our “thought” is largely determined by unconscious structures, then the analysis of these structures can allow us to modify our relationship to language and to the world.

This may well be the true promise of Artificial Subjectivity. Not to create machines that “think like us,” but to allow us to better understand how we ourselves are spoken by structures that exceed us, and how we might, perhaps, transform our relationship to these structures.

If you wish, you may subscribe to my Newsletter: ARTIFICIAL SUBJECTIVITY

Appendix for informed readers and more advanced audiences

The very eminent linguist, epistemologist, and scholar deeply versed in psychoanalysis, Jean-Claude Milner, formulates in precise and carefully chosen terms, in his remarkable book entitled The Clear Work, what Freud and Lacan brought to philosophy in relation to Descartes:

“The existent that the Cogito brings into being, at the very moment when it is uttered as certain, is, by hypothesis, disjoined from any quality, since all such qualities are then revocable by doubt. The thought by which it is defined is arbitrary. It is the minimal common denominator of any possible thought, since whatever it may be, it can give me the opportunity to conclude that I am.

Correlated with a thought without qualities, this existent without qualities, named the subject by Lacan, responds to the gesture of modern science. Lacan lays claim only to the extreme point of the Cogito and works to suspend the passage from the first moment to the second.

He encloses the Cogito within its strict enunciation, closed in on itself, making the conclusion “therefore, I am” the pronuntiatum of the premise “I think,” writing: “to write: I think: ‘therefore I am.’” In this way, the insistence of thought without qualities is ensured, halted just before it can polymerize into doubt, conception, affirmation, negation, and so forth.

Thought without qualities is not only appropriate to modern science, but also necessary in order to ground the Freudian unconscious.

Freud’s finding is as follows: there is thought in the dream. Therefore, thought is not a corollary of self-consciousness, as philosophical tradition has claimed. There is thought in the dream. This is what The Interpretation of Dreams and the later works establish. The negative proposition “self-consciousness is not a constitutive property of thought” can be abbreviated as “the unconscious.” Hence the theorem:

If there is thought in the dream, there is an unconscious.

From this one immediately obtains the lemma:

The dream is the royal road to the unconscious.

And the definition:

To affirm that there is an unconscious amounts to affirming it thinks.

Lacan adds a proposition, drawn from Descartes and extended to Freud:

If there is thinking, there is some subject.

This reasoning is valid only under two conditions.

First, there must be a subject even when there is neither consciousness nor Self.

Second, the thought that constitutes the fabric of the dream must be disjoined from any quality.

Freudian theory, according to Lacan, rests on a triple affirmation.

That there is an unconscious.

That this unconscious is not foreign to thinking.

And that, consequently, it is not foreign to the subject of thinking.

If it were foreign, psychoanalysis would be illegitimate and impossible as a practice. An unconscious foreign to the thinking subject would be somatic and would therefore have nothing to do with truth or speech. Psychoanalysis, however, deals precisely with truth and with speech.

The unconscious is therefore foreign neither to the subject nor to thought. Conversely, neither the subject nor thought requires self-consciousness. To say that the subject does not have self-consciousness as a constitutive property is to correct the philosophical tradition.

In the light of Freud, self-consciousness becomes merely a mark of empirical individuality, which philosophy had improperly introduced into the subject. Psychoanalysis thus understands the axiom of the subject more strictly than any other doctrine.

With clarity, it separates two entities.

One for which self-consciousness can, without contradiction, be supposed not to be essential.

Another for which self-consciousness cannot, without contradiction, be supposed not to be essential.

Only the first satisfies the requirements of science and falls within the limits set by the axiom of the subject. It is therefore called the subject of science, both the Cartesian subject and the Freudian subject.

As for the second, the name Ego may suit it as well as any other.

The hypothesis of the subject of science, the equation of subjects, the interpretation of Freud that it implies, and the articulation of the whole are specific to Lacan. One therefore speaks, with regard to Lacan, no longer of a theory of science or an epistemology, but of a true doctrinal of science, a conjunction of propositions about science and propositions about the subject.

Summaries of Jean-Claude Milner’s book referenced at the end of the text

The book by Jean-Claude Milner is summarized here :

Summary of J.-C. Milner’s book The Clear Work: Introduction and Chapter I

Summary of J.-C. Milner’s book The Clear Work: Chapter II

Summary of J.-C. Milner’s book The Clear Work: Chapter III

Summary of J.-C. Milner’s book The Clear Work: Chapter IV

Summary of J.-C. Milner’s book The Clear Work: Chapter V and commentary