In the US, decisions of the justice system are increasingly affected by algorithmic decision making systems. They are used in almost all US states and assist decision-makers in a wide range of decisions, e.g. about probation or the risk assessment for recidivists.
In a new article, computer scientists from the Algorithm Accountability Lab at the TUK, Prof. Katharina A. Zweig and Tobias Krafft, have joined forces with Georg Wenzelburger and put together a paper that discusses 1) how algorithms affect decision-making in the criminal justice system and 2) what difficulties and challenges arise from this from a political science perspective and theories of democracies. The paper entitled “On chances and risks of security related algorithmic decision making systems” has been accepted by the European Journal of Security Research and can be accessed here.
Certainly, this first paper is only the beginning of a longer collaboration between the Professorship for Policy Analysis and Political Economy with the Algorithm Accountability Lab. There is too much going on in this interesting field – especially from a public policy perspective. Algorithms are used in education policy (deciding in France about who is entitled to enroll in certain universities), in social policy (assisting employment agencies in their decisions about what candidate to propose a job offer), and, not least, in law and order policies. From a public policy perspective, it is important to first assess how such systems are used in different countries, how they are embedded in bureaucratic decision-making, and, most importantly, to analyze the policy process (including the various actors and interests involved in it) that led to the decision for or against the use of a certain algorithm.
Georg Wenzelburger
[photo by Thomas Koziel, provided by KLUFOS, University of Kaiserslautern]