Samenvatting

The promises of Big Data, predictive policing and artificial intelligence hold a key position in the public debate for quite some time now. Optimists tell that it is possible to predict where criminal events will occur before they take place. This would implicate a major shift towards a crime and insecurity preventive society, feeding on our cultural longing for a secure future. Therefore we give algorithms and deep learning access to more and more aspects of our lives. But how realistic and desirable is the application of Big Data techniques in the area of security?
In this article we put focus on the research question ‘In which way can Big Data and predictive policing support good governance of security?’, that has led our study. By exploring the central concepts, the processes behind them and their results in the domain of public security, we conclude that there are only rather disappointing results from the application of these techniques: crime and insecurity have not dropped when the police and other organizations turned to Big Data techniques. Instead, many negative side effects occurred. We search for explanations in six central academic critiques on the application of these techniques in the area of security.
We have found several ways to guaranty principles of good governance in the application of Big Data techniques, but these require a firm paradigm shift on Big Data in general. The heuristics of security professionals should not be overshadowed by technological promises: the professional should always be in the loop, must understand the way predictions come into existence and must be able to correct flaws and bugs of (semi-)automated decisions. We conclude that safeguarding public security must remain human work in which Big Data techniques can assist.