The boundary between professional and amateur artists became porous with the democratisation of creative tools. A third figure is now appearing, the machine that writes and makes images. I propose to think about it through the image of the child, which Alan Turing had also put forward as early as 1950, and to draw a practical consequence from it: what remains the province of human artists is the exploration of what does not yet have codes.
Artists as operators of a social fact
Art is a social fact among others. An artist may, from within their practice, claim to serve no purpose, but sharing uselessness is already a function, and one that plays a role in people’s lives. I therefore look at art from a sociological standpoint, as a set of functions whose operators are the artists, an approach I developed in the article Art and artists as society’s infrastructure.
If artists are the operators of this social fact, knowing who is an artist is not a sociologist’s quibble, since who occupies these functions depends on that designation. And this question has been reworked twice in about twenty years, first by the democratisation of creative tools, and now by machines that themselves produce texts and images.
Neither novices nor professionals
With mobile phones and social networks, the tools for producing images and stories have found their way into every pocket. On TikTok, sharing has become a form of creation in its own right. Through the play of trends, everyone is invited to take up, transform and extend what others have made, in a palimpsest logic where creating and sharing merge. Artistic legitimacy is built there on criteria of audience and constancy. If I decide to be an artist and publish regularly, being “consistent” as they say on the platform, I can build recognition for myself, without going through the institutions that, until then, designated artists.
The sociologist Patrice Flichy described this shift as early as 2010 in Le sacre de l’amateur (Seuil). The amateurs he observes are “neither novices nor professionals”; they are enthusiasts who draw real skills from their daily practice, to the point of rivalling the experts in their field, and he reads in this democratisation of skills the extension of the political and educational democratisation under way for two centuries. With the figure he calls the pro-am, the professional-amateur, the line that organised the art world becomes blurred. It has not disappeared, though; it has become far more complex than before, if indeed it was ever simple.
Two machines in dialogue do not go round in circles
In this recomposition, machines had so far occupied a stable place, that of capture and fabrication devices, reputedly a little dumb, the camera, the tape recorder. Wrongly reputed so, by the way. A given camera stamps its way of doing things onto images, stabilisers automate gestures, and the algorithms embedded in drone cameras carry a gaze of their own, to which I devoted part of my work on film art and drones. But those machines produced nothing unless a person pressed the trigger.
What is changing is that the machine now writes texts, makes images, proposes ideas. These productions are what they are, often imitative, often lacking originality. One will object that they merely respond to a human command, a prompt. Yet if two machines are put in dialogue, they do not go round in circles, they move forward, they prompt each other. The confrontation of points of view is even one of the technical keys of deep learning. In 2014 Ian Goodfellow proposed generative adversarial networks, in which a network that makes images confronts a network that judges them, each learning from the other. This creativity is not of the same nature as human creativity, and I examined what it does to the notion of authorship in the article The notion of authorship in the era of generative artificial intelligence. What interests me here is the social figure that follows from it. To the two figures whose boundary had already blurred, the professional artist and the amateur artist, a third one is added, the machine artist.
Vera Molnár, imaginary machine
This third boundary has a history, and it was first crossed in the other direction. In 1959, the painter Vera Molnár, who had lived in Paris since 1947, devised what she called her imaginary machine, simple programmes made of instructions and prohibitions she imposed on herself, and which she executed step by step, methodically, as would a computer she did not yet have. She co-founded the Art et Informatique group in 1967, gained access to a real computer in 1968, and became one of the pioneers of algorithmic art. She always held that the value of her works, or their lack of value, owed nothing to the machine that executed them. An artist had thus made herself a machine nearly ten years before touching one, and sixty years before machines produced images without her. The passage between artist and machine has long been worked on from the artists’ side; what is new in our decade is that it is now being crossed in the other direction.
Machines born of us
Machines are made by human beings, and they are imprinted with the cultures of those who make them, down to what their designers are not aware of putting into them. Then they take on a form of autonomy. The image that seems most accurate to me for thinking about this, and I offer it as an image, not as a demonstration, is that of children. Our children are born of us, they carry our genetic heritage and our cultures, they depended on us entirely, and yet they are not our product. They have their singular existence, they make their way in their own manner, and we have no mastery over them, whatever we may want to believe. Machines have, to my mind, that same nature. They are kinds of children, born of us, making their way.
Alan Turing put forward this image as early as 1950, in the last section of his article “Computing Machinery and Intelligence”, less often quoted than the imitation game that opens the same text. Rather than writing a programme that would simulate the mind of an adult, he proposes there, let us try to produce one that simulates a child’s, and then subject it to an appropriate course of education. The child’s brain, he writes, is presumably something like a fresh notebook bought from the stationer’s, very little mechanism and lots of blank sheets. And he adds two remarks that already describe our current machines. The teacher of a learning machine, he warns, will remain largely ignorant of what is going on inside it, and these machines will surprise their designers, which was his answer to Ada Lovelace’s objection, for whom a machine could only do what we knew how to order it to do. Today’s deep learning is the realisation of that programme, with child-machines educated on our texts and images, opaque to their own teachers, and capable of surprising us. What makes a child impossible to master, the fact of being born of us without being our product, is found again in these machines.
Funding the improbable
It remains to say what this figure changes in the work of human artists. Pre-codified creation, the kind whose rules are already known, can be imitated, and machines will produce it better and better; I developed what this implies for commercial cinema in the article Artificial intelligence is an opportunity for free and experimental cinema. Entertainment, in the sense of what diverts us from working on ourselves, will be very well made by machines, and it will entertain us very well.
The machine imitates codes. Where codes do not yet exist, what remains are experiments, experiences to attempt, projects that seem absurd, for which no criterion has yet been established; I devoted the article The question of criteria to this absence of criteria. To move creation forward, and audiovisual creation in particular, we therefore need to break new ground, that is, to fund the most improbable, the most fragile, the least mastered projects, because that is where things get invented. If we believe we can master what is to come, we lose, exactly like the parent who would believe they master their child.
And we need to be there. If we leave the fields of creation to the industrial players who operate generative artificial intelligences, we let them draw the future. I believe it is important to be there in a thousand ways, by creating with these tools, by creating without them but in conscious relation to them, by taking them as subjects of works, by putting them in relation within our works, whether visibly or covertly. Every artist has their own way of doing things; what I hold to is consciousness. Children have been born of us who write and make images. The place of human artists is not to compete with them on what can be codified; it is to take the side paths, where, to my mind, we will reinvent our humanity, through art.
