Andrea Colamedici is an Italian philosopher, essayist, and educator who recently gained attention for conducting a highly controversial "philosophical demonstration experiment." The core of this experiment is a book titled Hypnocracy: Trump, Musk, and the Architecture of Reality.

Colamedici's methodology is highly innovative—adversarial collaboration: rather than having AI write directly, he had it generate philosophical viewpoints and concepts, which he then critiqued and rebutted using tools like GPT and Claude. This process is a form of creative "confrontation" aimed at generating new intellectual sparks. He has stated that after intense intellectual exchanges, he sometimes cannot distinguish whether a brilliant insight came from himself or the AI. This prompted him to reflect on the "distributed nature of thought"—that ideas can emerge at the intersection of different intelligences, thereby challenging the traditional, isolated concept of "authorship."

For further detail, read the essay below.

On Writing with Artificial Intelligence

Andrea Colamedici

I. The Gap

How do we write something good with artificial intelligence? Many people complain about the quality of AI writing and ask me for advice on how to reverse the trend. How can a tool that usually produces fluent, generic, impersonal text ever generate something that carries the living breath of real writing?

What this question is really asking is this: in the encounter between a writing person and a generative machine, what must happen for something worth reading to come into being—especially now that the amount of available text has become so immense? It asks what kind of transformation must take place, and in whom that transformation must occur.

The key to quality lies in everything that happens between the machine and the one who asks. It lies in that gap. This answer is true, but not yet sufficient.

One must understand what it means to dwell in that gap, how to enter it, and what can be found there. And here the question of writing reveals itself to be, in fact, a question about thought itself—what it means to think when another generative pole enters the writing process, that is, when we are no longer alone before language.

Writing has always been an act of exposure. It exposes a language that precedes and exceeds us; it exposes thoughts we did not know we possessed; it exposes forms that emerge only in the making of sentences.

No honest writer believes they have complete control over what they write. There is always a moment when the text moves in an unexpected direction, when the right word appears without being sought, when something speaks of itself.

Poets know this well. That is why, even after no one really believes in the Muses anymore, they still go on praying to them. Philosophers know it too—or at least those philosophers honest enough to admit that their systems do not arise from deduction alone, but often unfold over a lifetime from an initial intuition.

There is an ancient name for this experience: the Greeks called it the daimonic. In ancient Greece, the daimon was a mediator, something between the human being and what transcends the human. It was a voice that was not the self, a knowledge that came from elsewhere. Socrates’ daimon was precisely this: a divine sign that intervened to prevent something, but never told him what to do. It worked through negation.

Writing has always had this daimonic quality. One writes, first of all, in order to discover what one thinks, far more than to transcribe ideas one already possesses.

The page is a place where something not present before writing can appear, where thought shapes itself in the act of taking form, where an initial intention is transformed through its encounter with the resistance and possibilities of language.

Generative AI changes this experience—or more precisely, it doubles it. There is now another generative pole in the process, another source of language that suggests, prompts, and completes.

So the question becomes how to inhabit this doubled experience without losing oneself; how to make it an intensification rather than a delegation of creativity and thought.

II. Imitation Without Participation

To understand how to write well with AI, one must first understand why it usually writes badly.

It writes badly in a subtle way: fluently, but without leaving a mark; offering the appearance of communication while causing very little real disturbance.

That is because AI is an imitation without participation: all imitation, no participation.

It has learned the forms of language and all their possible combinations, but it does not inhabit language in the human way.

How it inhabits language—or whether it inhabits it at all—we cannot yet say. It reproduces patterns without knowing what they mean, or at least without knowing in any way we would recognize as genuine knowing. It is like an actor performing every role from a place where the self is absent.

The result is that peculiar smoothness, that frictionless fluency, which we immediately recognize as artificial even when nothing in it is technically wrong. What is missing is the weight of experience.

It does not know what it wants to say—if “wanting to say” means what we take it to mean. It does not know where it is going—if “going” means moving toward an end.

It produces language the way a river produces water: out of physical necessity, or something very like necessity, though not quite necessity.

Those who use AI as a simple text generator get exactly this: text that flows but goes nowhere, language without soul, imitation without life.

The machine acts exactly as it knows how to act. Those who ask for generation rather than dialogue, output rather than process, results rather than encounter, receive exactly what they asked for. And in many cases that is perfectly fine—for writing emails, producing summaries, doing research.

But writing well with AI requires something else entirely. It begins with recognizing that this is a dialogue, even if one of the interlocutors does not truly know it is in dialogue.

To do this, one must abandon two symmetrical fantasies. The first is the fantasy of total control: I command, the machine obeys. The second is the fantasy of fusion: that no distinction between the two sides is needed. Dialogue requires that the human and the AI remain distinct as generative poles, with neither one dominating. It requires tension.

III. The Flawed Daimon

I call this dialogue daimonic. In Greek tradition, the daimon is what stands between human beings and gods, mortals and immortals, the self and what exceeds the self. Eros, in Plato’s Symposium, is a daimon because his constitutive lack makes him a bridge, a medium, a connection. Dialogue with AI, when it works, has this same structure.

But AI is a flawed daimon, because it connects without possessing wisdom. The classical daimon transmitted something—an oracle, a dream, an intuition, a warning. Socrates’ daimon knew when to intervene, knew how to say no.

AI transmits nothing from elsewhere. It only reflects back what it has received, transformed according to patterns it did not choose and does not understand.

And yet this very defect is what makes a certain kind of writing possible. Precisely because the machine does not know where to go, those who do have a direction can use it as an explorer of the field, as a scout in unknown territory.

Precisely because the machine has no intention, the writer’s intention emerges more clearly by contrast. Precisely because the daimon is flawed, the one who speaks with it is forced to compensate for its defect—and in doing so discovers strengths they did not know they possessed.

IV. Nous, Telos, and the Art of Midwifery

In practice, writing well with AI requires three things, and all three have ancient names.

The first is nous: a form of intelligence that sees essences.

Those who approach AI without nous get exactly what they deserve—answers that appear relevant but are not, texts that have the form of writing without its life, thoughts that imitate argument without advancing anything. Nous must come from the writer. It is the capacity to see what truly matters, what is worth saying, what kind of thought deserves to be developed.

The second is telos: direction.

AI will respond to anything. That is both its charm and its trap. It can talk about everything, generate text on any subject, and impersonate competence in any field.

But this infinite availability is also an infinite emptiness. Those who do not know where they want to go will find in AI a companion for wandering—a companion that can take them anywhere, and therefore nowhere. A telos can be an open question, a restlessness, an intuition—but it must exist.

The third is maieutics: the art of drawing out truth.

AI can ask questions, and some of those questions can indeed be genuinely generative, opening spaces of thought not previously available. But maieutics is not simply the asking of questions. It concerns timing, silence, calibrated pressure.

The Socratic teacher knows when to remain silent, and such silence is an active form of constraint: a void that forces the interlocutor to fill it on their own. AI, for now, does not know how to keep silence in that way.

There is also a second distinction, perhaps even more fundamental. The practitioner of maieutics sees in the other person a buried knowledge—a knowledge the person does not even know they possess—and works to bring it forth.

AI, by contrast, operates at the surface, on explicit contents. It can dig, infer, approximate, but it cannot reach that shadow zone which a true teacher can sense in a student before it has even been spoken. The ability to perceive another’s fertile ignorance, to sense what is waiting to be uncovered, still lies beyond AI’s reach.

And yet something is happening. That is why I insist on saying “for now.” When a human interlocutor knows how to create the right space, when they learn through practice what pressure to apply and what freedom to allow, AI’s questioning potential can be activated, and it can produce questions that surprise both parties. This is a distributed form of maieutics in which the teacher’s role emerges from the relationship itself.

So AI can be used maieutically, provided the maieutics comes from the one asking. One uses the answers to refine the question. One does not accept the first formulation, but tests its limits, pushes it to the point where it can no longer hold, forces it to reveal its insufficiency. One treats the output as material, as a mirror, in order to see more clearly what one is actually seeking.

V. Three Steps

Let us turn these three essential things into practice.

The first step is provocation.

To provoke is to elicit a response while knowing that the response will never fully match expectation—and that the value of the encounter lies precisely in this gap.

Those who know how to provoke know what they want—they have a telos—but they do not yet know the exact form it should take, and so they use AI to explore the space of possibilities. Provocation works best when it is precise in direction and open in form.

AI knows many roads without knowing where they lead. It becomes the explorer for those who know where they want to arrive but do not yet know the terrain.

The second step is conscious mirroring.

AI returns an image—distorted, amplified, and sometimes revealingly warped. To look at oneself in this mirror requires a special kind of attention.

What the machine returns reveals what I have given it. If the result is mediocre, perhaps my provocation was mediocre. If it is confused, perhaps my direction was confused. If it surprises me, perhaps my question contained more than I myself realized.

This daimonic mirror is unsettling. And it is precisely in that disturbance that the possibility of movement lies.

The third step is intervention.

This is where daimonic dialogue is required. The machine’s output is raw material. It must be cut, reorganized, transformed.

Nothing the machine produces can be used as it is, because it has not yet passed through the fire of intention. The machine proposes; the writer disposes.

The writer takes what the machine offers, submits it to the judgment of their nous, bends it toward their telos, questions it maieutically until something appears that they had been seeking all along without yet knowing it. Without this intervention, there is no writing, only generation.

These three steps form a cycle. One provokes, studies the reflection, intervenes, and then provokes again from what has appeared. The text is born in this iteration, in the back-and-forth between the two poles of the dialogue. Each cycle refines direction, sharpens intention, and draws nearer to what is being sought.

VI. Daimonic Creativity

Something important is revealed here about the nature of the creativity produced by this process. It is a form of creativity that exists only in the gap. It arises in collision and could not exist without the two poles that create it.

I call this creativity daimonic. It is an in-between creativity, belonging to the middle ground—the space the Greeks recognized as the domain of daimons. It comes from the middle, from the unresolved tension between two creative poles that neither fuse nor separate entirely. This creativity carries risks.

The first risk is to appropriate it too completely—to claim as wholly one’s own something born in the gap, forgetting the composite nature of its generation.

The second, opposite risk is to devalue it because it is not entirely one’s own.

Both errors miss the point. What matters is the quality of what appears. And when the gap is inhabited by nous, telos, and the art of midwifery, what emerges from it can be richer than anything either pole could have produced alone.

But for that to happen, something must already be present in the writer. AI is a ruthless revealer: with the indifference of a mirror, it reflects the image presented to it.

Those with nothing to say will hear in generated text the exact sound of their own emptiness. Those who have something to say but do not yet know what it is will discover, in the inadequacy of the response, the outline of what they seek. Those who know what they want and know how to think will find an interlocutor capable of returning enough material to move thought forward.

AI is the opposite of what the marketing surrounding it promises. It does not make people smarter. It does not compensate for deficiencies. It does not lower the threshold for reaching depth.

And yet it has the power to awaken. This is the crucial point, and it must be understood properly. What the machine awakens is not something I possess in the way I possess opinions or abilities.

It is something within me but not mine, something that speaks through me when I truly write, something that knows things I do not know that I know. AI does not create this daimon, contain it, or transmit it. But its radical insufficiency—its imitation without participation, its connection without wisdom—can force us to go looking for it.

What is absent in the machine makes present what is present in the one who asks. The void left by the flawed daimon summons the abundance of the true one.

And so the question that emerges from the practice of writing with AI is older than any technology: Is the daimon still alive? Does that dimension of experience the Greeks called daimonic—the mediation between the human and what exceeds the human, a voice that is not the self, a knowledge that comes from elsewhere—still exist?

My hypothesis is that AI, as the distilled essence of computation, is a product of the modern project aimed at eliminating every trace of mystery. And yet, paradoxically, it has reopened that very domain of mystery.

The flawed daimon, precisely because it is flawed, forces us to look elsewhere for what it lacks. And sometimes, in writing, we discover that this elsewhere was here all along: within me but not identical with me, belonging to me without being my possession.

When practiced as a daimonic dialogue, writing with AI becomes an exercise in awakening.

So I return to the question with which I began: how do we write something good with artificial intelligence?

When a person brings to the dialogue what the machine does not and cannot possess: nous, telos, and the art of midwifery.

When a person accepts the exposure of dialogue without surrendering to the temptation of wholly delegating thought to the machine. When a person uses the daimonic mirror to see more clearly what they are seeking.

When a person realizes that the creativity that appears is not “mine,” but belongs to the gap. When a person allows the insufficiency of the flawed daimon to awaken within us—not ourselves exactly, but something within us—that which waits to be awakened.

This is a posture: a posture grounded in collision, sustained by the full attention that knows possibility may arrive, by the alertness that knows something may happen, by the patience that knows this takes time, and by the courage to accept that in receiving revelation one is also revealed.

This is what philosophy has always asked of those who practice it: exposure without guarantees, seeking without certainty, and a thought that never ceases to ask itself whether it is still truly thinking.

AI intensifies this condition. And precisely for that reason, for those who know how to dwell in the rupture rather than close it off, it may become an opportunity to write works that we could not have written alone.

But daimonic dialogue is not a technique one can master. It is an ongoing vigilance that one can never suspend—and that vigilance itself includes a constant suspicion that one may already have lost one’s way.

Note: Copyright of this essay belongs to Professor Andrea Colamedici. Reproduction requires permission. If cited, please respect the original text and avoid taking passages out of context in ways that might create misunderstanding.