Risk Assessment Instruments and the Erasure of Discretionary Judgment: Rethinking Human Agency in Technology-Mediated Criminal Justice and Correctional Administration

Sheilakate Vanderpuye

Abstract: The increasing use of risk assessment instruments (RAIs) and algorithmic decision-support tools across criminal justice and human correctional systems has generated concerns about fairness, transparency and the boundaries needed to keep such tools in check. Unfortunately, little theoretical attention has been given to this a subtler institutional risk: the gradual erasure of discretionary judgment among criminal justice professionals who operate within algorithmically structured environments over long periods. Drawing on organizational theory, rationality frameworks, and literature on automation bias, this paper argues that even low-stakes or nominally advisory risk scoring tools can alter the cognitive and professional dispositions of their users over time. This phenomenon is defined here as discretionary attrition, and it poses long-term risks to the quality, legitimacy, and humanity of decision-making in criminal justice. This paper presents a framework for understanding how discretionary attrition happens in correctional and sentencing settings. It also explains when this issue becomes more common and suggests ways to help maintain human decision-making in criminal justice systems that use algorithms.

Keywords: risk assessment instruments, discretionary judgment, automation bias, criminal justice, correctional policy, human agency, sentencing