For many proponents of the use of pretrial risk assessments, the hope is they provide an evidence-based counter to racial bias in the criminal justice system. However, as has become apparent with the more widespread adoption of the tools, they can also reproduce the very racial bias they were intended to disrupt. This paper grapples with the question of whether it is possible to address the problematic aspects of risk technologies without abandoning their use.
The paper considers the evidence supporting both sides of the debate: that risk assessments can be used to render racial inequities more transparent (and hence capable of correction), and that they can further entrench preexisting disparities. The paper also examines why people of color fare worse in risk assessment scoring. Ultimately, the authors contend the real debate over the use of risk assessments is not one related to math or statistics, but values. The paper concludes with a number of concrete suggestions for improving the performance of assessments: policy strategies concerning the application of risk tools, and technical strategies to revise the tools themselves to maximize equity.