Adults, phone in hand, deplore the time children spend in front of screens. Teaching materials critical of artificial intelligence are produced with its help, without anyone saying so, or even thinking about it. I propose to call “screen discourse” the sincere discourse, borrowed from common opinion, which occupies in a person the place where their own practice would need to be thought through, and I propose four concrete tools for dealing with these contradictions.
The truth of what we say is not only what is said
Linguists distinguish the utterance, which is what is said, from the enunciation, which is the act of saying it, performed by someone, addressed to someone, in a given situation. Émile Benveniste, in “The Formal Apparatus of Enunciation” (1970, reprinted in Problèmes de linguistique générale II, Gallimard, 1974), defines enunciation as the putting into operation of language through an individual act of use. Before being a content, speech is an act, and this act has a place, a moment, circumstances, a body. From this I draw a consequence that goes beyond linguistics: to judge the truth of what someone says, it is not enough to examine the utterance, one has to look at the situation of enunciation, that is, the acts that go with it.
In my work in media and film education, I regularly come across two situations where this gap is plain to see:
- Parents worry about the time their children spend in front of screens, and lecture them about it, their own phone in hand, and are the first to answer notifications in the middle of meals. Teachers who ban phones from the classroom have their own lying on the desk in front of them.
- There are teaching materials for the youngest classes that teach that artificial intelligence is merely a tool, one that will never replace knowledge, experience or human relationships, while the images in these materials, their graphic design, sometimes their outline, were developed in dialogue with an artificial intelligence, without this being stated anywhere, and often without the people who designed the material having even thought of stating it. The text is a warning against the medium, and the object is made of that medium. The philosophers Karl-Otto Apel and Jürgen Habermas call performative contradiction a statement contradicted by the very act of stating it; here the contradiction is not only logical, it is material, inscribed in the object.
The contradiction in itself is not, to my mind, the heart of the problem. We all live with our contradictions, and in the article The paradox of digital criticism I developed the idea that the problem is not having them, but not seeing them. What interests me here is more nuanced: the contradiction entirely escapes the awareness of people who are subtle, cultivated and in good faith, with neither stupidity nor cynicism involved. This precise phenomenon has no name. I would like to propose one.
From screen memory to screen discourse
In 1899, Sigmund Freud published a short text, “Screen Memories” (“Über Deckerinnerungen”), in which he describes childhood memories of surprising sharpness for perfectly insignificant scenes. He proposes to understand these memories by their function: their very sharpness is enough to arouse suspicion, for each of them stands in the place of another memory, more charged, that cannot be kept in mind as it is, and which thus remains out of reach. The person who recounts a screen memory is not lying, the scene most often really took place, and it is because the memory is real that the mask is so opaque.
I propose to do for discourse what Freud did for memory. There are screen discourses just as there are screen memories. I call screen discourse a discourse held sincerely, generally taken over from common opinion, whose function, for the person who holds it, is to occupy the place where their own experience would need to be thought through.
I see four features in it:
- First, we do not invent it, we take it over. The screen discourse exists before the person who utters it; it circulates in the media, in institutions, in conversations; it is available, ready to use, and that is why we take it over so readily.
- Second, it is sincere. The person who states that artificial intelligence will never replace human relationships believes it at the moment of saying it, and that is the whole difference from hypocrisy, which presupposes that one knows.
- In holding it, we protect ourselves on three fronts: we spare ourselves the examination of our own practice, which would take time and courage; we shelter from the judgement of the group, since repeating common opinion costs nothing socially, whereas we expose ourselves as soon as we speak in our own name; and we preserve the position of authority we occupy, as educator, parent or expert, which is harder to hold once we expose our own practice.
- Finally, and this is the least expected feature, it is by speaking that we create the screen. If we kept silent, the void would show; by speaking, we fill it. Speaking learnedly about a subject then becomes the surest way of not thinking about it.
In the case of screens and artificial intelligences, the word fits doubly well, and I did not look for it: the discourse held about screens itself acts as a screen.
A rich experience, a poor discourse
Screen discourse also differs from cognitive dissonance, which I discussed in the article Cognitive dissonance and Artificial Intelligence. Dissonance presupposes that the two contradictory terms meet in the same consciousness, and that this meeting produces a discomfort which the person then seeks to reduce. In the situations I am describing here, there is no discomfort at all. The two registers coexist without ever meeting. The same person will gladly recount, with pleasure and with subtlety, their own uses of artificial intelligence, the way they worked out a graphic design by conversing with a machine, the surprises they had along the way, the precautions they learnt to take, their lucid analyses of how their relatives use artificial intelligence. And this same person, in the same conversation, holds on artificial intelligence in general a discourse that is poor, normative and false. Their experience is nuanced; their discourse is not. Their experience belongs to them; their discourse, they received ready-made, without realising it.
This asymmetry between the richness of experience and the poverty of discourse is the mark of screen discourse. To open up awareness, I simply propose asking “and you, how do you do it?”. When this question is asked, the screen discourse comes undone, not because the person is caught out, but because their answer, as soon as it describes their real practice, is finer, more nuanced and truer than the discourse they were holding a minute earlier.
In the article The social cost of coherence, I proposed three mechanisms for understanding those areas where commitment no longer operates, without our seeing it: the radical monopoly of tools, face-work in conversations, and the welding of commitment to identity. With the concept of screen discourse, I would like to name the sentences through which these mechanisms pass. Something had to occupy the place left empty by the work of thought that does not get done; common opinion is full of ready-made sentences, and it is by repeating them that we exempt ourselves, without ever having decided to, or even noticed it, from questioning our own experience. And in the article Conscious compromise with artificial intelligence, I argued that what is at stake, faced with our contradictions, is not to achieve perfect coherence but to maintain an attention to what we are going through. Screen discourse settles in wherever that attention is missing.
Two messages at once: Bateson’s double bind
In 1956, the anthropologist Gregory Bateson and his team, Donald Jackson, Jay Haley and John Weakland, published “Toward a Theory of Schizophrenia” (reprinted in Steps to an Ecology of Mind, 1972). There they describe what they call the double bind: a person simultaneously receives two contradictory messages, sent at two different logical levels, from someone on whom they depend, and it is impossible for them to comment on the contradiction. This last point is decisive in Bateson: the double bind only becomes destructive when the contradiction cannot be commented on, and the solution he proposes is metacommunication, that is, talking together about the way we talk to each other.
Children to whom screen discourses about screens and artificial intelligences are addressed are caught in a structure of this kind. They receive the utterance: beware of screens, a machine will never replace your friends. And they receive the situation of enunciation: adults with phones in hand, parents consulting an artificial intelligence at the table, teaching materials produced with the very tools they are being taught to distrust. I leave aside Bateson’s clinical hypothesis, which concerned schizophrenia; I retain the structure, and above all the condition that the contradiction cannot be commented on. For children see the contradiction. They know what their parents do. In these situations, the real learning is not about screens: children learn there that the contradiction between discourses and practices is not to be commented on, and so they learn, by immersion, the art of later holding discourses detached from their own practices. Screen discourse is passed on, and it is passed on through the situation of enunciation, never through the utterance.
Keeping writing for the scribes
It remains for me to propose a hypothesis about where these discourses come from, for their very sincerity calls for an explanation: why do we so readily take over sentences that do not come from our experience? The anthropologist Jack Goody showed, in The Domestication of the Savage Mind (Cambridge University Press, 1977), that writing is not a mere recording of speech but a technology of the intellect: those who practise it no longer think in the same way, because the list, the table and the formula make possible mental operations that are impossible orally. In Literacy in Traditional Societies (Cambridge University Press, 1968), Goody described societies of restricted literacy, where this technology remains in the hands of scribes and clerks who draw their power from it. The history of technologies of the intellect is also a history of powers. Those who hold them control access to them, and readily address, to those they keep out, a discourse of protection.
I propose, as a hypothesis and without being able to establish it, to place within this continuity a part of the educational severity towards children’s digital uses. Artificial intelligence is a technology of the intellect of new power. Adults use it, draw considerable professional and personal advantages from it, and hold to children a discourse through which they implicitly reserve its legitimate use for themselves. I attribute this intention to no one: it is the very nature of screen discourse not to be calculated. But this ancient structure, technology for us, morality for them, seems to me one of the reasons why we take over these discourses so easily.
Another clue supports me in this reading: the normative maxims addressed to children, if one listens carefully, are not addressed to them. A sentence such as “a machine can answer you, but it is your brain that has to think” brings nothing to children, who go through situations with their emotions long before relating them to principles; it is the sentence adults say to themselves, to convince themselves they have done their job of educating. The stated recipient is the child, but the real recipient is the person speaking. I see in it one of the purest forms of screen discourse: speech whose function is spent in comforting the person who utters it.
Making contradiction a subject of conversation
There remains the question that matters most to me: what to do? I do not believe in the demand for perfect coherence, whose price I described in the article on the social cost of coherence; no one keeps to it, and demanding it mostly produces concealment. The path I take from Bateson is more modest and more practicable: making it possible to talk about contradictions. I propose four tools, which require no particular resources:
- Say how it was made. People who produce a document about a medium say whether they used that medium to produce it, and how. A three-line production note is enough. With it, the object changes status: from a mute double bind, it becomes a support for conversation, and children can ask why these images, who chose these colours, what the machine proposed and what the humans kept.
- Include your own practice in the subject. One does not teach a medium from above. Teachers and parents who talk about screens with children also describe their own uses, their pleasures, their difficulty in switching off. By exposing ourselves in this way, we lose a little of the position of authority, and we gain a situation of enunciation in tune with the utterance, which is the condition for being believed.
- Start with “and you, how do you do it?”. In a meeting, a training session, a school council, before collecting general opinions about screens or artificial intelligence, each person first describes their own practice. When we start from what we actually do, we can no longer repeat ready-made sentences, because they no longer match what we have just described. It is an elementary democratic device, and the most effective I know against ready-made discourse.
- Get help to see. My own screen discourses are, by definition, the ones I do not see. I proposed, in that same article on the social cost of coherence, that we offer one another a view of our blind spots. With screen discourse, this exchange has a precise object: the sentences I repeat without their coming from my experience. Asking someone I trust to point them out to me is probably the most demanding tool on this list, and the most useful.
A last word on the use of this concept. The most tempting thing is to use it against others, since we see screen discourse in others as easily as we fail to see it in ourselves. But by denouncing, we undo nothing: the person denounced clings to their discourse, which protects them, and we cheaply fabricate for ourselves one more screen discourse, that of lucidity, which allows us to stop looking at our own practices. I propose the opposite use, turned towards oneself and practised with others: finding who, around me, will agree to point out my ready-made sentences, and to whom I can render the same service. This takes a certain courage, but this courage, that is to say this risk-taking, seems to me the cornerstone of the educator’s role.
