Enhancing the Integrity of the Sentencing Process through the Use of Artificial Intelligence

Abstract

Abstract The likelihood that a defendant will reoffend is one of the key considerations that informs sentencing outcomes. It is relevant to the aims of community protection, specific deterrence, and rehabilitation. Despite this, empirical data establishes that judges make poor predictions regarding whether defendants will reoffend. Algorithms are being increasingly used in some parts of the United States to determine the risk that offenders will commit other crimes. This practice has come under considerable criticism for a range of reasons. This chapter examines the persuasiveness of criticisms of sentencing algorithms, discusses the relative limitations of both human and computational decision-making, and argues that properly designed artificial intelligence systems can significantly enhance the integrity of the sentencing process.