The term predictive justice refers to the use of AI algorithms and techniques to forecast
legal outcomes based on historical data and patterns.
Our study explores the concept of predictive justice, which involves utilizing AI algorithms
and techniques to forecast legal outcomes based on historical data and patterns. The application
of predictive justice will be facilitated by the Council of State's decision to open up its decisions,
starting from March 2022. The Council of State now provides open access to court decisions for
all three levels of French administrative jurisdiction : administrative tribunals, administrative
courts of appeal, and the Council of State. The motivation behind this study lies in the timeconsuming
nature of legal proceedings and the need to automate repetitive tasks. To address
this, empirical research and AI methodologies, such as machine learning and natural language
processing, are employed to analyze a collected database of administrative court decisions. To
establish a foundation for this research, a comprehensive literature review identifies the commonly
used machine learning models for decision prediction in the European Court of Human Rights
and the Supreme Court of the United States.
The primary objective of this study is to conduct a comparative analysis between the French
judicial system and other legal systems worldwide. To achieve this, the study applies the most
commonly used prediction models found in the literature of other countries. Our approach is
novel because predictive models have not been extensively applied to the open data of the
French Council of State until now. By undertaking this analysis, the study aims to determine the
level at which a prediction model can accurately forecast a decision in a French administrative
tribunal. Ultimately, our study seeks to demonstrate that prediction models applied in law cannot
be generalized universally, as what may yield results in other contexts may not be applicable
to French administrative decisions. This analysis can benefit citizens by providing them with
an idea of the potential outcome of their case and can also save time for legal professionals by
providing an overview of the proceedings.
Furthermore, our study highlights the application of information extraction techniques to
extract relevant information necessary for decision-making within each prediction model. By
harnessing the capabilities of AI, legal professionals can gain a comprehensive overview of cases,
enabling them to make more informed decisions efficiently.
The findings presented in this article contribute to understanding the potential benefits of
integrating AI into the legal system, specifically in the field of predictive justice.