Computer-Assisted Sentencing in Germany and the EU

• In Event: Understanding Criminal Proceedings

Abstract

For some time now, an intense debate has been ongoing regarding the opportunities and limitations of computer-assisted sentencing. One hypothesis posits that such systems have the potential to enhance consistency in judicial decision-making and, consequently, promote greater fairness in the imposition of sanctions by addressing structural deficiencies in the current sanctioning process, such as statistical regional disparities and cognitive biases in human decision-making. Recent technological advancements, particularly in computing power and machine learning, have facilitated numerous practical implementations and experimental applications. Moreover, a considerable number of theoretical models have been proposed. The legal assessment of these systems cannot be conducted in a generalized manner but must instead account for the diversity of technical methodologies and their respective implications for judicial autonomy. This study introduces a systematic classification of computer-assisted sentencing systems into 21 distinct variants, based on their technical design and degree of influence on judicial decision-making. This taxonomy serves as a foundation for a subsequent legal analysis, which differentiates between these categories with particular consideration given to the European Union’s Artificial Intelligence Act. The examination will delineate the current legal constraints applicable to each category and, based on these findings, assess the necessity of further legislative interventions.

Author

• Hauke Bock, Humboldt-University Berlin