Presaje Institute Symposium – June 25, 2019

We attended a symposium organized by the Presaje Institute, an independent think tank, “dedicated to the analysis of the complex relationships between economics, law and justice”. Several judges, lawyers, academics and researchers of different nationalities gathered on this occasion in the premises of the Court of Cassation, around the theme of AI and the hopes and concerns it raises within the judicial world.

The title question “A judicial world enhanced by artificial intelligence? “introduces the questioning to which the two round tables were invited to discuss.

Bruno PIREYRE, President of the Chamber, introduced the debate with a historical overview of the AI, weighing the expected benefits of a more detailed analysis which should enhance legal certainty against the risks of a loss of freedom of decision for actors in the judicial world. Like any new technology, AI must be approached without false fear, but with caution, without naivety. Among the initiatives contributing to the rational use of AI, he mentioned the transparency of algorithms, mentioned in the 2017 Cadiet report on the open data of court decisions, as well as the CEPEJ’s European ethical charter on the use of artificial intelligence in judicial systems and their environment.
Bruno PIREYRE’s speech is online on the website of the Court of Cassation.

The AI, a new judicial actor?

A first round table discussion, moderated by Thomas CASSUTO, Magistrate, Vice-President of the PRESAJE Institute, answered this question with several testimonies explaining the place occupied today by the AI:

  • Nathalie NAVON-SOUSSAN, lawyer, testified about her use of predictive tools to prepare her pleadings based on the probability of success of the cases she submits for analysis. The lawyer, like any professional, needs tools, and this type of tool allows him to advise his clients more accurately.
  • Fabien TARISSAN, CNRS researcher, professor at the ENS, explained the mechanisms for learning deep learning. For machines to be able to learn from the data, it is necessary that the data be initially described by a human being. The issue of bias based on insufficient or poorly trained learning data is often cited, with the example of the COMPASS algorithm used in the US to estimate prisoners’ risk of recidivism.
  • Yannick MENECEUR, Magistrate seconded to the Council of Europe, presented the 5 principles of the CEPEJ’s ethical charter:
    • Respect for fundamental rights,
    • Non-discrimination,
    • Quality and safety,
    • Transparency, neutrality and intellectual integrity,
    • User control.
  • Soraya AMRANI MEKKI, Director of the Judicial, Amicable and Digital Justice Axis at the University of Paris Nanterre, presented the difficulties of using AI in terms of alternative dispute resolution, as defined by the Law for the Reform of the Justice System of 23 March 2019. The said law authorizes and specifies the operation of online services providing conciliation, mediation or arbitration services. It creates confusion between the different alternative dispute resolution methods of amicable settlement and arbitration. The “amicable” game would be distorted by the proposal of an algorithmic decision, to be taken or left….. And how can we ensure that informed consent is obtained on such a platform? Finally, the possibility of certification of these platforms remains to be clarified. Many questions remain unanswered.
  • Adrien BASDEVANT, lawyer, author of “l’empire des données”, recalling Laurence LESSIG’s famous article “code is law”, denounced the competition between computer, legal and ethical codes.

Towards automated justice?

Faced with this inevitable introduction of AI into the judicial world, the second round table, also moderated by Thomas CASSUTO, raised the question of the advantages and risks of automation.

  • Mrs Kaï HÄRMAND, from the Estonian Ministry of Justice, testified about the deployment in her country of automated procedures for the settlement of minor offences, reducing the workload of judges and registrars, allowing them to focus on more complex cases. Expected benefits: reduction in the time required to process small claims and reduction in the cost of justice.
  • Michel MORVAN, President of Cosmo Tech and the System X Institute of Technological Research, pointed out the fundamental difference between the functioning of the human brain and that of artificial intelligence: the human brain knows how to recognize, and in this it operates in a black box mode, just like an AI. But, in addition, he knows how to reason, link, infer, project, what an AI can’t do from nothing. Data alone is not enough, as many believe, to make an AI work. He also needs a model that provides him with reasoning. It is in modelling that the difficulty in the judicial field lies and this is where Mr. MORVAN situates the main risk, much more than in the biases resulting from the selection of data.
  • Christiane FERAL-SCHUHL, President of the National Council of Bars, reminded us that, in the continental legal system, laws change frequently, thus challenging previous jurisprudence. The use of predictive justice tools would thus be more appropriate for the Common Law system, where case law is the dominant factor. The risk of a shift to a commmon law system and the establishment of rules of law is not excluded. The lawyer’s human role must be preserved, as well as his or her freedom to deconstruct case law, by placing each case in its own particular context. In order to limit the risk of data bias, Me FERAL-SCHUL insists on the need to have a complete, representative and controlled case law database. To this end, the Court of Cassation and the National Council of Bars have requested in a joint declaration that the open data of judicial decisions be placed under the exclusive responsibility of the Court of Cassation.
  • Christophe REGNARD, Magistrate, Honorary President of the International Union of Magistrates also pointed out the changing nature of French law and the risks of using AI: frozen, dehumanized justice, judicial errors induced by the probabilistic nature of the recommendations made by the AI.

    Conclusion

    Paul NEMITZ, Principal Advisor, DG Justice European Commission concluded this superb and very dense morning with a general encouragement to France and Europe to invest in the mastery of these technologies, in accordance with their ethical and democratic values.