Justice remains one of the last areas that has not yet fully experienced the digital revolution. Industry, finance and previously culture, particularly publishing and broadcasting, have experienced the upheaval of the world's data entry. So now we hear more and more every day about "predictive justice" or "legaltech".
Algorithms hold the promise of more fluid, more efficient regulation. Probability rather than judgment. The computer code rather than the legal code. Governance by calculation rather than by law. However, the root of these risks has already been pointed out by Alexis de Tocqueville: "The notion of government is becoming simpler: numbers alone are the legislation and the Law. All politics is reduced to a question of arithmetic."
What should then be the added value of justice in relation to machines? Which thread should it follow, in this labyrinth of original, binary architecture, composed of 0 and 1; where justice is necessarily plural, work of interpretation, charged with symbolism? In an era of artificial intelligence fuelled by massive data, algorithmic governance would easily convince us of the time saved and the lower cost of these new public, judicial or more generally social decision-making processes. But that would be a disregard of the essential, because it is the way in which a decision is taken and motivated - in particular a court decision - that determines its legitimacy and the responsibility of the person who makes it. The law cannot avoid the analysis of the complexity of the act in order to judge an individual. Justice must be accountable.
This implies the possibility to understand the algorithms, to be able to question them, or even to contest them. The controversy surrounding Parcoursup [French platform completely dematerialised that manages all of the enrolment procedures for an institute of higher education], the national procedure for distributing post-baccalaureate admissions, bears witness to this.
Thus, in order not to get lost in the maze of digital technology and to preserve the rule of law, technological choices must be subject to democratic debate. In order not to succumb to the overthrow of power by the digital age, in order not to suffer a "Coup Data", we must preserve the ability to deliberate and make our voice heard as a citizen, i.e. as a "subject of law".
This requires the initial and above all ongoing training of all. Citizens, magistrates and politicians must develop a critical digital culture to be able to use and compare statistical and algorithmic models. Otherwise, we will dilute our responsibility by relying on the diagnosis of these black boxes, and not be able to verify the criteria used, which may be biased or discriminatory. Education, therefore, above all, so as not to maintain a new form of voluntary servitude. What would otherwise be the meaning of this evolution that would drown our individualities in a quantifiable mass where any possibility of questioning our intentions, of questioning the models establishing our desires, by transforming us into a sum of raw, objective data, without personality? Maintaining and defending the irreducible singularity of each and every case is the task of justice in the digital age.