Artificial Intelligence and Law: Computational Models of Argument and Explainability Solutions

As a result of the urgent need for explainable decisions in AI, several techniques have been developed over time. This thesis reviews the most meaningful approaches, with an emphasis on their application in the legal system, emphasizing argumentation as the most suitable explainability method. It also includes a section that compares deep learning to supervised classical AI, highlighting the differences and implications for explainability in legal AI systems; and a section about the legal framework concerning explainability in Artificial Intelligence in order to ensure that future systems provide reliable, accurate and impartial decision support, with convincing arguments to back up machine judgements.