Everyone has heard of the outrage provoked by the discovery that Facebook databases were used to influence the Brexit decision in England and the election of Donald Trump in the United States.
But what exactly is it about? It is rather simple, alas. Until 2014, Facebook’s terms of use stipulated that a user’s consent also gave access to the data of their “friends” (friends) without those friends’ consent.
A British researcher then obtained, in the most legal way possible, the consent of 270,000 Facebook users to supposedly have them take a psychological test. Relying simply on Facebook’s rules, he thereby gained access at the same time to the data of all the “friends” of those who had responded to him. As each Facebook user has on average 200 friends, this British researcher found himself holding data concerning nearly 50 million Americans. He then passed this data on to a “friendly” company, which used it for political purposes.
The computing power of computers now makes it possible to sort citizens into categories not only according to their age, social class, political and religious concerns, but also according to their culinary tastes, clothing preferences, cultural and sporting interests, holiday destinations, and so on. This sorting by category then makes it possible to send each person emails, Facebook messages or tweets perfectly targeted at their interests. In this way it becomes easy to influence a large number of people by using, for each one, an argument that touches them intimately, as if it came from a “friend” who knows them well and shares their concerns. That robots are generating these messages is of course unknown to those who receive them. So you may receive a message from someone who says they discovered you through social networks and share your interest in a particular film star or pastime, and after saying a few words about your shared passion, slip in a remark about the social or political state of the country, express their worries (which are obviously yours, since the robot is programmed to know them), or even tell you of their hope for a change that you also have in mind.
The aim of these targeted messages is not to make you change your mind, but to confirm each person in their preconceptions through a contact they believe to be close, so that they become less receptive to rational arguments. Everyone’s positions become caricatured, debate hardens, and no one listens any more to the arguments of politicians, because they have “well understood”, thanks to the messages of all those who think like them (in reality the robots), that they are right.
The question this raises is simple: is the free social network model, which rests on the commercial exploitation of its members’ personal data, not in contradiction both with ethics and with the possibility of democratic functioning? Indeed, in a democracy, political messages disseminated by lobbies are explicitly identified as such. In France, it is in fact one of the roles of the CSA (the audiovisual regulator) to ensure this. But the targeted messages of robots on networks escape the CSA, and they can seriously influence citizens without their knowledge, endangering the democratic logic based on debate, curiosity about the other’s point of view and the exchange of ideas.
S.T.
Serge Tisseron is a psychiatrist, doctor of psychology with accreditation to supervise research, member of the Académie des technologies, and associate researcher at Université Paris VII Denis Diderot (CRPMS).
