AI-Powered Investigations: From Data Analysis to an Automated Approach Toward Investigative Uncertainty

Criminal law might sometimes be perceived to be at the margins of the automation process that involves increasing sectors of our society. While the expansion of automated-driven cars is by now an established fact, the technological upgrade of criminal justice mechanisms still tends to evocate sci-fi images and Minority-report-style dystopian scenarios to the non-specialists. However, the high variety of applications that can already be counted in this domain, and the acceleration of this process in the last few years, clearly speaks for a tangible expansion of AI technology also in the field. In particular, against some more-established scenarios, especially related to phenomena of so-called predictive policing, Multi-Agent Systems (MAS) open today the perspective of a much deeper involvement of automated technologies in the very shaping of investigative proceedings. The chapter offers an analysis of such potential, both with respect to their possible contribution to the efficiency of investigations (in their preventive and repressive dimension), and to avoid or reduce certain negative biases typical of “purely human” investigation processes, first of all, the tunnel vision effect.