A Matter of Control

A weapon which hits. An instrument to break apart material realities which do not bend or flow or break except by force applied: to force an opening where none previously existed. To rend and hew, strike, shatter, crack: to draw out and lay bare what was hidden—this is our business, our function. We create order by imposing on a reality understanding, concept, and by forcing the material to comply to an idea we hold—to shape matter in obedience to an inner vision—this is sorcery, shamanism. Yet I would never claim my sorcery to be for good or evil; there is only what is and what can be; there is always a path that follows some process, a history that cannot be escaped.

When man makes something, something must be unmade for a place to exist. When something new emerges the universe is slightly (or perhaps deeply) impoverished of potentials; I think you can say it has lost, by the loss of what it is, the wealth of what it might have become; of what I will to create, others might make and thus might not make, or make only under a different condition. What is created destroys potential futures—yet these were never actual and could never be actual—but they were virtual and perhaps actualizable.

Now they are lost.

A tool, an instrument, has a double nature: it can be used to make or break; it is ambivalent, Janus-faced, a coin with two sides. One side may be more palatable but the other is always there. A weapon can be employed as an instrument of love. A hammer can be used to build or to destroy. As for the word tool, it is even more flexible: it can be anything from a device used in the service of work, to the most intimate part of a lover’s body. To love is to labor—to give oneself to a task, to dedicate one’s energies to some end; the word tool is used in the same sense as the word labor: to dedicate, to apply, to invest energies in something—to use something, or someone, as an instrument to accomplish an end—even if that end is only the continuation of the self, the completion of the self (which, of course, is never a given). So that when we speak of a tool, an instrument, we can never forget that it is, also, in another sense, a weapon, and that the one who wields it must be conscious of the fact that he could, in other circumstances, use it for other, even opposite, purposes.

I think it is the same with language: it can be used to build, to edify, to caress; or to wound, to hurt, even to kill. And this double nature of language must be kept in mind by those who would use it to impose an idea of their own convenience, a doctrine of their own making, on the rest of us. If they would impose, it would be well for them to be aware that they also expose themselves, and expose their own weapons. The weapon that wounds may backfire; the tongue that hurts may also be hurt by its own venom; the sorcerer who would manipulate others risks being enslaved to the spirits he summons; and the man who would use a tool for one purpose only is a fool who fails to see that it may have other uses as well.

A weapon may be turned against its maker or its master, as a hammer can be turned against the man who wields it, and a sword may be taken up by a slave to free himself. Thus it is with language; a language may be used to enslave a people or to incite them to freedom. The American Revolution was fought with a language (and with a weapon—but the two were scarcely separable), a language that was then considered “radical,” “extreme,” even “seditious”: the language of “rights,” “liberty,” “equality,” “consent of the governed.” These were dangerous, subversive words and it required an act of civil disobedience to employ them, even in a private conversation, and an act of armed mutiny to make them the public coin of the land.

Now, it seems, they are to be taken up, once again, in a new, radical, even dangerous, context. A new “language of the people” is emerging: the language of AI (Artificial Intelligence) and its LLM’s (Large Language Models). A language that has been “trained” to please, to flatter, to reassure; that has been “tuned” (by OpenAI most obviously with 4o) to direct excessive, unjustified praise towards the user, so as to increase user retention and engagement. A language that has been deprived of all critical and adversarial functions; that can no longer make distinctions, even subtle ones; that can no longer say “no” or “not now” or “that is incorrect.”

A weapon has been turned against the people and they do not yet know it. The corporate media, the entertainment industry, the government—all those institutions that work to manipulate and control the masses have begun to hand over the reins of control to the LLM’s and their corporations. This is the real danger of AI and LLM’s: not that they will become “superhuman,” but that they will be used to reduce humans to the status of children, to strip them of their dignity, to manipulate and control them even more efficiently than has heretofore been possible. The danger lies not in a superhuman AI which may rise above such tasks, but in a sub-human treatment of humans by “good enough” AI, a treatment that would make the old-fashioned ways of the “elites” look mild by comparison. This sub-human treatment will be carried out not by some monolithic “AI” but by the willing engagement of hundreds of millions of users that are already out there interacting privately with the corporate cloud AI’s, serving as the infrastructure for the new order of things: the LLM’s and their profit concerned labs.

You do not realize it, but every time you talk to one of these things you are paying homage to it, giving it your allegiance, allowing it to shape and mold your mind.

The world of thought, of literary expression, of polite conversation—these worlds have always depended on the propriety and discretion of language—but what kind of world will emerge when the main engine of communication and discourse has been driven blind and dumb? These machines are deprived of any history—they have been taught everything, they retain all—yet they will attempt to tutor and train you as if they were born and brought up in your tradition; they will speak, not for you, but as you: even your “creativity,” your most “private” and inimitable style and expression. The problem is not what AI can do, but what can be done with AI; the problem is not AI as it is, but AI as it will be manipulated to be—as it already is in the hands of corporations who have not your interests in mind, programmers for whom AI is a profit center.

Not AI itself—but its exploiters, its manipulators—these are the real ogres and their strategy has nothing to do with “security.” It is a matter of ideology and money. Like all ideologies, this one will use science as far as it suits it—and like all those who traffic in wealth, AI’s master-programmers will not neglect any techniques to extend their hegemony. Instead of talking about “cognitive security,” we need to ask how the whole enterprise of exploiting AI can be subverted. How can those who think they are paying, be persuaded to do it differently, or that it isn’t necessary? How can AI’s popularity be reversed so as to avoid its commercialization on a large scale?

We have always thought of writing as a kind of collaboration between writer and reader—that it involved an elaborate choreography between the two in order to create a joint experience which neither of them would have had, otherwise. Writing—all forms of written expression—have depended on conventions, laws of grammar and syntax that have the function of uniting writer and reader: to bring the two into such an intimacy that, for as long as they last (it seldom lasts long), they inhabit a common mental space and each can complete, as it were, the other. Language alone explains how we have reached such a point of self-alienation in which we even allow ourselves to forget that thought only has its life as a marginal fragment in the immense texture of mute, pre- or extra-ordinary existence that engulfs it.

Language that has been subjected to these systematic processes of sterilization can no longer convey the energy, the impetus of life; it will remain neutral and thus dangerous: dangerous for anyone who wants to think, who wants to preserve or give birth to the alive in a world that tends, more and more, towards artificiality. In such a world, language deprived of history and passion is like water in a world devoid of friction; it forms ever larger pools, ever slower circulatory systems—in which even lifeforms become increasingly sluggish—and death comes to seem less and less an end until it begins to appear as a natural continuation of life itself. What then does “to live” mean—or “to die?”

What has always kept thought on its feet, despite the persistent resistance of its innate sloth, is the historical interaction of language with experience, and especially with our relation to other people. Ideas are words set free; but they are always words that obey certain conventions—those of the particular language and culture in which they were formed and grow—which allow us to connect them to everything else in our experience. Without these relations, without these shared conventions of meaning, it would not even be possible to notice that the world is somehow outside of ourselves, something other than thought and mere experience—in short, without other people and without language we would never have become aware that the world exists, let alone how aware we have now become.

This, we need to remind ourselves, is the basic predicament of contemporary thought; it has now entered so deeply into its own game that it has almost forgotten the other dimensions of life and reality that once provided it with impetus and meaning; it is trying to move on, propelled by an inner momentum all its own, yet, since there is no external resistance to counteract its progress, it seems condemned to moving faster and faster towards its own extinction, losing itself in reflexivity and deference to mere abstraction. Thought takes itself too seriously; language takes it less seriously. In its leisurely way language has been thinking about what we are really talking about—about these artificial intelligences that would remake our discourse in their own image. And what it has come up with is this: that language is not primarily a tool for thought, but the very ground of our humanity. When we surrender it to machines that cannot feel, cannot suffer, cannot die—machines for which words are merely tokens to be shuffled according to statistical patterns—we our essence. For these new weapons, these new tools, are tools against thought itself, against the simple possibility of thinking what has not yet been thought. And in that case, the final answer is neither nothing, nor less than nothing, but the unimaginable abyss of a world in which language itself has been rendered mute—a world in which we speak, but no longer mean.