Paper's abstract

Elise Mouriesse, What Transparency for Predictive Justice Algorithms?
The contribution is based on the observation that there are currently few transparency requirements regarding predictive justice algorithms made available to judicial and administrative judges to help them adopt court decisions, despite proposals made in that direction. This paper considers why such requirements should be imposed and how they could be realized. First by explaining the reasons why such requirements would be desirable, recalling the numerous risks inherent in algorithms, especially when making them available to judges for the purpose of predictive justice, and by demonstrating that applying transparency requirements could counterbalance these risks. Then it presents the potential concretization of these requirements, first by strengthening their potential legal bases. It then builds on the requirements of administrative transparency applicable to administrations that base individual decisions on algorithms to propose a modeling of transparency requirements applicable to judges with regard to the degree of use of algorithms in the adoption of court decisions, with due regard for the prerogatives of the magistrates and the guarantees they enjoy.

Key Words : transparency algorithm
t. 60, 2018: p. 125-145

---