Predictive justice is one of the most hotly debated applications of AI. In it, artificial intelligence could be used as a tool whose purpose is to predict court decisions thanks to an algorithm that mines legal and case-law data. Predictive methods developed for the field are gaining ground.

With the development in France of the Internet, artificial intelligence and digital tools, in particular the launch of legal decision “open data”, justice must now face new challenges. Thus, using algorithms to process the most repetitive and simple cases could reduce, among other things, the time spent on research using comparable matters of fact or of law. According to the deputy chair of the Conseil d’Etat (State Council), speaking at a conference in February 2018, “predictive justice would help focus judges’ time on cases in which their expertise is of greater added value”.

Predictive justice also offers economic players a unique tool as regards anticipating and harmonising court decisions. Still according to the deputy chair of the French Conseil d’Etat, “open data and algorithms should together foster access to law and fairness before the courts, and encourage the stabilisation, harmonisation and convergence of case-law. There are, of course, more traditional methods to achieve this goal that are just as effective. But unity and consistency of case-law can, it is true, improve using current and future technological developments”.

Predictive justice does have its critics as regards its operation and the decisions made. Its quality and performance are under careful scrutiny, especially in an era that has become dependent on technologies and information. Technological progress must not hinder a judge’s work and above all, for those persons answerable to the law, must not hinder access to justice. It is often stated that, “although the predictability of law is necessary, it must not freeze case-law “.

It is therefore essential, for those players hoping to enjoy this technology, to organise themselves so that they may anticipate its limitations. Faced with these innovations as regards the application of the law, the fundamental principles of justice must nevertheless remain untouched and not be altered in any way by the predictive algorithm. Since there is a risk of mimicry in the use of these tools, they must be used with the knowledge that judges retain their freedom to assess and decide independently.

Thus, companies hoping to anticipate their legal risks and/or measure the legal impact of breaching regulations, for instance, must be prepared for external data that tools of predictive justice are, as yet, unable to fully appreciate: a judge’s authority to decide.