The illusion of thinking: from Turing to the unconscious

From the very beginning, research on artificial intelligence has been rooted in the areas of interfaces and simulation. In his first paper on the subject in the mid-1930s, Turing did not propose to build an intelligent machine as such, but rather one capable of simulating the effects of human intelligence. Questions about what intelligence is, or what it means to think, were never central to his research program. Instead, Turing was driven by other questions, such as how to imitate intelligence and achieve results similar to those we attribute to our own thinking. The core of early twentieth-century cybernetics was focused on the how, the similarity, and the imitation.

At least initially, Turing’s program was even more limited. The English logician’s quest was not to imitate our thinking but rather to imitate the very special ability of our thinking, which consists of counting and performing elementary mathematical calculations, such as addition and subtraction. These are the operations that Blaise Pascal and Charles Babbage had found ways to perform in their famous machines in the seventeenth and nineteenth centuries. But Pascal’s and Babbage’s machines were machines in the literal sense, made of wood, steel, and interlocking gears. Based on certain mathematical relationships between the myriad wheels and cylinders they were made of, they could even show the result of an operation on what we would call a display. Turing’s machine was an entirely different kind of machine. It was a logical machine, and at least at first, it was a purely theoretical machine. Something that one day may be built based on the details of its operation being defined then, but at the time, there was no thought of really building it.

The interface of another interface

The flow of Turing’s article leaves no doubt. It goes something like this. Imagine a tape upon which you can store some data. Imagine a stylus capable of reading that data. Imagine a mechanism capable of moving the tape by the number of positions specified by the number read. What we have before us is not a machine but an interface between us and another machine that is currently not on the agenda. The machine described by Turing is not a machine but something that computer science, when it emerged a few decades later, called a program, or software. So the question is, what is software? And where should we imagine this other machine we have been discussing?

Consider the work of a film editor. Their job is to effectively render an action by stringing together, within an appropriate syntax, specific sequences that a director has chosen to shoot. None of us would ever mistake those sequences for action, and none of us would believe that the editing of those sequences is the same thing as the event the movie is about. Something similar happens with the Turing machine. The Turing machine performs the operations that logic, the philosophy of mathematics, and what in the second half of the twentieth century became structuralist linguistics and semiotics have classified as the activity we call thinking. In short, the operations of this machine mimic something that is already of the order of mimicry; they perform something that was already intended to reproduce our thinking by entirely different means, and they enact something that was more or less consciously limited to simulating our mental functioning by stringing together certain operations to which our thinking had been preemptively reduced. 

It is a description of what we do when we think, and only in this way does it become an actual specification of what a machine would have to do to achieve a result comparable to ours. Thus, a software program is not primarily an intermediate language that allows the machine to do what we do, reducing it to the scope of its clumsy and crude means. Rather, it’s an intermediate language that enables us to make an assumption about what we do when we think, reducing us primarily to the scope of ourselves, making us primarily the machine of ourselves. In other words, software is a hypothetical mirror that humans build around themselves. 

The two discoveries of the unconscious 

A first way to describe the discovery that gave rise to psychoanalysis is to say that while we are used to thinking that there is a certain subjectivity, that there is an “I” that thinks, understands, desires, and meets its peers, psychoanalysis shows that something quite different is at work. This very different something is the unconscious; so-called subjectivity is only a superficial effect of the dimension unknown to subjectivity. Mistakes, dreams, and missed actions are no longer random events that occur to unreasonably interrupt the continuity of our conscious lives, but rather clues to that other dimension that is always at work. Clues that the unconscious would make known in the translational language imposed on them by their emergence into consciousness, just as the shapes we imagine at the bottom of the river come to us deformed by the thickness of the water they must pass through to reach us.

Perhaps Freud imagines the unconscious in a still-romantic way, as a repository of repressed traumas and unknown attachments that find a way to break through in times of weakness and distraction of our consciousness. But his follower, Jacques Lacan, translated the discovery of the unconscious into terms more suited to twentieth-century sensibilities. Every subject, says Lacan, is driven by a linguistic machine whose gears turn ceaselessly, producing not only our dreams or mistakes but our desires, our deepest beliefs, our way of seeing the world and interacting with others. And not only that. This linguistic machine, like a machine made of tiny cogs and belts, revolves around a signifier that is something like its immovable engine. You are my champ, you are mother’s love, you just happen to be here. Everybody has their master signifier, as Lacan calls it.

The idiot signifier

But the point is that every master signifier is dumb, empty, and idiotic. This is the most interesting thing if we want to use cybernetics and psychoanalysis together to flush out that enigmatic point where their trajectories intersect as they sketch the space of our contemporary subjectivity.

Lacan’s discovery is not, in fact, that we all have our master signifier, and that this master signifier is the firebrand that the word of our father or mother has left on the blank page of our unconscious. This is true, but Freud already knew this. Lacan’s discovery, in which he shows himself to be fully Thuringian, is that this key signifier, which is also at the root of every other effect of meaning, is, as far as he is concerned, devoid of any meaning. 

How can it be explained, how can it be said, how can it be made intelligible to us, if not through its effects? For example, how can you say what it means for me to be my mother’s favorite, or to be my father’s champion, or to have come here by chance, if not by using a whole world, a whole language, a whole life of experiences that were shaped first and foremost by the fact that I felt my mother’s favorite, or by being convinced that I came here by chance, or by being called my father’s champion? It would be like trying to figure out what color my sunglasses are without ever taking them off for a second in my entire life. I can swear that they are the same color as this piece of sky or this tree leaf, but I have only ever seen this piece of sky or this tree leaf filtered through these steadfast glasses.

How to use a short circuit well

In other words, the big question we find at the intersection of the legacies of these two twentieth-century giants is not whether machines can think but whether human beings, indeed them, really think. And further, in what sense do we say they think? And what do we ultimately mean when we say they think? What idea of thinking do we have, how did we get it, and why, and with what useful consequences and what harm? 

Because, precisely, everything we say about thinking, about our own intelligence or that of our machines, about human and artificial minds, is played out behind the fact that we are thinking it, behind the fact that we use some kind of intelligence to define it, behind the fact that some mind is at work to describe the mind we call human or artificial. Some mind, mind you. Not our mind and not the human mind. Even this thing we call our mind is an idea; even this other thing we call the human mind is also a construct. 

It is a short-circuit, but if we inhabit it properly, it is an instructive short-circuit. Ethically constructive, politically constructive. Constructive, that is, of an ethics of our knowledge, of a politics of our technologies. Like dreams, which never speak of the people we dream of but of how we dream them and how we are, likewise machines and models of humans speak of us, of how we dream of ourselves, and how we are or think we are. This “we” we speak about also speaks primarily about the fact that we talk about it, much more than about the “we” we think speaks. Whatever interface we choose will, in short, promise to ease the path to a certain elsewhere, but deeper down, it will return us to ourselves in the form of a question. Who are we, exactly?