Virginia’s Use of Supervised Machine Learning to Code Judicial Departures from Sentencing Guidelines Recommendations

Abstract This article focuses on the Virginia Criminal Sentencing Commission’s development of state-specific artificial intelligence (AI) and machine learning (ML) processes that encourage acceptance of the technology and improve outcomes. Addressed are concerns on the use of AI and ML to enhance the manual coding of written departure reasons provided by judges and the impact on job functions. The article describes the technical process used to develop ML programming specific to Virginia while also bolstering the acceptance of the technology by staff. The article details how the programs and automated processes were developed to code judges’ reasons for going above or below the Guidelines recommendation.