Abstract

The essay discusses the subject of Artificial Intelligence (AI) algorithms and the legal principles governing this matter. The functioning of the algorithms Turochamp, Stockfish, AlphaZero, used in the game of chess, is described in detail. Based on these examples, the differences between deterministic, non-strictly deterministic algorithms and self-training AI algorithms are analysed, up to the topic of opacity and black box, typical of deep-learning systems. Finally, the issue of the transparency and explicability required by jurisprudence for the use of AI algorithms in the public sphere and the compatibility of these legal principles with the functioning of these new technologies are addressed