The Rise of Algorithms in the Criminal Justice System

The use of algorithms, especially intelligent ones, promises to foster innovation, efficiency, and effectiveness in policing and criminal justice. In an attempt to satisfy an increased desire for security and zero tolerance in this field, jurisdictions all over the globe are starting to deploy algorithms to support processes in police work, criminal legal proceedings, and ultimately, the execution of imposed sanctions. Methods of predictive policing have evoked an especially sensational new paradigm, relying on the assumption that crimes are subject to statistical regularities. Algorithms promise to calculate offenses at time X and place Y, or predict at what probability person Z is likely to commit a crime. With such tools rapidly being disseminating, a “smart criminal justice” has begun to emerge.

An article by Monika Simmler and Giulia Canova

The age of algorithms is dawning in the criminal justice system. Algorithms are being designed to predict crimes, automate legal proceedings and forecast recidivism. The increasing relevance of algorithms in this field is not only depicted in headlines such as “Wrongfully accused by an algorithm” and “AI is sending people to jail­–and getting it wrong,” but also by a rising number of scientific contributions dedicated to the use of algorithm-based systems in policework and the criminal justice system (e.g., Završnik, Hayes et al., Peeters and Schuilenburg).

Against this background, a research project dedicated to the use of (intelligent) algorithms in the criminal justice system has been initiated at the University of St. Gallen. The project explores the use of algorithms in the Swiss criminal justice system, and asks how far evolved such “smartness” already is. In so doing, it has and continues to analyse the related institutional, psychological, social, and legal consequences and identifies avenues for further research.

Drawing on research on smart government (e.g., Gil-Garcia et al.), the notion of smart criminal justice encompasses the use of technology in the criminal justice system based on algorithmic decision-making and the collection, analysis, and processing of big data. However, in the sensitive area of criminal justice, the implementation of advanced algorithms comes with fundamental questions regarding their usefulness, legitimacy, and lawfulness. Not every use of technology is intelligent or smart per se. To qualify as smart, governmental use of algorithms has to adhere to normative principles such as equality and non-discrimination. This applies to the criminal justice system in particular. In sum, smart criminal justice describes a criminal justice system that in addition to deploying smart technology is also construed in a smart manner.

Since the concrete extent of the use of algorithms in the Swiss criminal justice system was largely unknown, an empirical study was conducted to provide much-needed foundational knowledge. This research explored the implementation of algorithms in different areas of criminal justice and revealed that several algorithm-based systems are already in use in police work and the penal system. However, most of the technology can hardly be classified as “intelligent,” and accordingly exhibits low degrees of automation and technical autonomy. Still, the study showed that algorithm-based tools are of considerable importance in various areas of the Swiss criminal justice system.

A paradigm shift is currently observable in police work, brought about by predictive policing moving the focus from prosecution to crime prevention. On the one hand, predictive policing tools that are designed to identify potentially dangerous individuals and prevent them from offending are increasingly in use. This identification often results in lists or databases with profiles of seemingly dangerous persons. Subsequently, these persons are given special care by the police and interventions often take place. Conversely, spatiotemporal predictive policing applications are used to identify possible crime scenes and the times of their occurrence. Such applications exhibit a certain level of algorithmic intelligence, as they model risk areas from past data (e.g., on burglaries), and the dataset is constantly expanding. Furthermore, the Swiss police uses data structuring instruments that visualize connections between offenses; however, these tools are not of very high technical complexity, nor are they self-learning. In criminal investigations, data collection and analysis techniques are well established. In the digital age they are increasingly improved (such as by advanced datamining tools and face-recognition technology). Lastly, in the penal system, an algorithmic triage tool is deployed that screens convicted offenders for risk of relapse and automatically determines if a person requires further clarification regarding correctional measures or enforcement decisions. Additional tools, mainly in the form of (electronic) checklists, are used for forensic-psychiatric assessments to evaluate the behaviour and risk of recidivism of a defendant or convicted person.

Figure 1. Categories of smart criminal justice.

Aside from revealing that algorithms are indeed being used to a growing extent, this study allowed for general conclusions to be drawn regarding technology use in criminal justice. It became clear that the implementation of algorithms was often triggered by key criminal events and media pressure, and is in line with a general shift from the repressive function of criminal justice to an increasingly preventive orientation. The study also shed light on the psychological components accompanying the use of algorithms. It was revealed that individuals who had acquired algorithmic tools generally had a positive attitude towards them, while individuals deciding against acquisition expressed scepticism. The concrete application of the tools is furthermore affected by an interplay between the user’s gut feeling and the algorithmic results. Although the latter promises ‘objectivity’, the respondents admitted that when in doubt, they listened to their personal feelings. Moreover, the study disclosed uncertainties about the legal foundation of predictive policing instruments. Also, the mostly absent (independent) evaluations of the effectiveness of such tools was addressed. The fact that such (legal and scientific) uncertainties arise only after the implementation and use of algorithms is cause for concern.

In sum, the study identified different challenges regarding the use of algorithms, even though the algorithms currently in use in Switzerland are not particularly complex (yet). More advanced tools may bring more severe consequences. Next to challenges regarding concrete implementation and optimized human-machine interaction, questions regarding fundamental rights, data protection, and criminal procedural law will shape the scientific and legal discourse in the future.

A well-founded discussion of the use of intelligent technology in policing and criminal justice is crucial in the age of algorithms. The book project Smart Criminal Justice (open access in German here) is dedicated to this discourse. The collected articles comprehensively discuss the risks and potential of digitization for criminal justice and address some of the most urgent questions regarding the path towards a smart criminal justice.

Published under licence CC BY-NC-ND. 


  • Monika Simmler, Prof. Dr., is an assistant professor of criminal law, law of criminal procedure and criminology and the director of the Competence Center for Criminal Law and Criminology at the University of St. Gallen. Her research focuses on the impact of modern technology on the attribution of criminal responsibility, as well as on criminal justice in general. She is particularly interested in the interface between (preventive) police work and criminal prosecution.

  • Giulia Canova is enrolled in the Masters Program of Law and Economics at the University of St. Gallen and is research assistant at the Competence Center for Criminal Law and Criminology at the University of St. Gallen. In her research, she works primarily at the intersection of IT and criminal law, as well as IT and law enforcement.