Artificial Intelligence in Criminal Justice Settings:: Where should be the limits of Artificial Intelligence in legal decision-making? Should an AI device make a decision about human justice?

Abstract

The application of Artificial Intelligence (AI) systems for high-stakes decision making is currently out for debate. In the Criminal Justice System, it can provide great benefits as well as aggravate systematic biases and introduce unprecedented ones. Hence, should artificial devices be involved in the decision-making process? And if the answer is affirmative, where should be the limits of that involvement? To answer these questions, this dissertation examines two popular risk assessment tools currently in use in the United States, LS and COMPAS, to discuss the differences between a traditional and an actuarial instrument that rely on computerized algorithms. Further analysis of the later is done in relation with the Fairness, Accountability, Transparency and Ethics (FATE) perspective to be implemented in any technology involving AI. Although the future of AI is uncertain, the ignorance with respect to so many aspects of this kind of innovative methods demand further research on how to make the best use of the several opportunities that it brings.