What about AI in criminal intelligence? From predictive policing to AI perspectives

Abstract

Predictive policing is more and more developed around the world. TV-shows and fictions such as ‘the minority report’ or ‘Person of Interest’ spread a pre-crime effect that is, nevertheless, very different from reality. Many law enforcement bodies develop predictive analysis to find new opportunities against crime and it is generally dedicated to patrols. The Gendarmerie Nationale in France carried out through the concept of criminal intelligence a way to provide relevant information to describe, understand and foresee crime at different scales: operational, tactic and strategic. The aim is to upgrade the process of decision-making. Because crime is nor a random process neither a deterministic process, some features exist to characterise it. Obviously, it is very difficult and probably not possible to identify all features linked to crime evolution or criminal behaviour. Nevertheless, some characteristics are not so complicated to model in a formal mathematical structure. So, in the age of Big Data, applications of predictive analysis can be overtaken by artificial intelligence (AI). It is very developed in fields like medicine, finance or transportation and could on one hand provide new perspectives to fight crime but also on the other hand raise questions for future. Who will be the next organisation able to assure the best way to anticipate crime and criminal behaviour? AI could be defined as the capacity of a computer to model human reasoning. A grand challenge is opened for law enforcement but only if they are able to adapt their way of working to this new era. The scope of this paper is to describe the French development in predictive analysis and to open the potential use of artificial intelligence in different area of criminal intelligence without avoiding the risk of its new development.