Julian V. Roberts (University of Oxford - Centre for Criminology) and Mathis Schwarze (Free University of Berlin - Department of Law) have published "Reconciling Artificial and Human Intelligence: Supplementing Not Supplanting the Sentencing Judge" on SSRN. Here is the abstract:
How can Artificial Intelligence (AI) best contribute to the sentencing process? Should it supplement or supplant human decision-making? This essay considers the options. The essay begins by briefly noting some reasons why a machine-learned sentence represents an unconvincing alternative to judicial decision-making. This section touches on the central feature of sentencing in common law countries: the sentencing hearing, which constitutes a distinct phase of the criminal process. According to some scholars, AI can replace the human judge. We reject this view. Instead, we propose a number of ways that AI may supplement judicial decision-making. Computer-assisted information systems are the most-discussed approach and several have been implemented going back as far as the 1980s. We discuss the problems in constructing an adequate database and the benefits of an AI-derived Advisory Sentence. We also explore other ways in which AI may improve the sentencing process, for example, by detecting bias and unprincipled departures from principled sentencing more effectively than conventional means. We also note the unexplored ways in which AI may improve the guidance provided to courts.
Comments