By: SHAMENA ANWAR, JOHN ENGBERG, ISAAC M. OPPER, LEAH DION

The use of artificial intelligence (AI) methods to aid with decisionmaking in the criminal justice system has widely expanded in recent years with the increased use of risk assessments. Nowhere has this shift been more dramatic than in the widespread adoption of AI-enabled risk assessment tools to aid in pretrial detention decisions.

Despite the promise of pretrial risk assessment tools, the ways in which these tools have been implemented has limited potential progress. The vast majority of jurisdictions that have implemented these tools have essentially provided these risk assessment recommendations to judges in an advisory manner and generally cannot require judges to follow the recommendations when making their pretrial release decisions. Studies indicate that judges frequently ignore the recommendations of the risk assessment instrument; as a result, the adoption of these risk assessment tools has not had much impact on reducing the use of monetary bail and pretrial detention.

In this report, the authors investigate the factors that are predictive of whether judges follow risk assessment recommendations and identify the impacts to pretrial detention, public safety, and racial disparities when judges follow the recommendations more often.

RAND Research - Published Sep. 5, 2024

By Rafael A. Mangual 

After enacting a sweeping bail reform, New York lawmakers have drawn the ire of constituents who are troubled by the many stories of repeat and serious offenders—some with violent criminal histories—being returned to the street following their arrests. In the state’s biggest city, the public’s growing concerns are buttressed by brow-raising, if preliminary, crime data, amplifying calls for amending or repealing the bail reform. The operative provisions of New York’s bail reform severely limit judicial discretion in pretrial release decisions, increasing the number of pretrial defendants who are being released, often without conditions and without allowing judges to consider the risk that a defendant poses to the public. New York is now the only state that does not allow judges to consider public safety in any pretrial release decisions. This brief begins with an overview of New York’s pre-2020 bail law and the reforms that took effect on January 1. It then highlights the reform’s shortfalls and ends by proposing three changes intended to address the public’s legitimate safety concerns while preserving the spirit of the reform effort and addressing some of the inequities and inefficiencies inherent in a system that is heavily reliant on the use of monetary pretrial release conditions. The proposed changes include: • Empowering judges to assess the public safety risk posed by pretrial defendants, and setting out a process that allows them to detain dangerous or chronic offenders; • Allowing judges to revoke or amend release decisions in response to a pretrial defendant’s rearrest; and • In the intermediate term, setting aside additional funds or diverting existing funds to reduce the time a defendant stands to spend in jail if remanded to pretrial detention.  

New York: Manhattan Institute, 2020. 14p.

By Liz Komar and Nicole D. Porter

After 50 years of mass incarceration, the United States faces a reckoning. While crime is far below its peak in the early 1990s,1 the country continues to struggle with an unacceptable amount of gun violence.2 Meanwhile, the drug war harms too many Americans and has failed to prevent fatal overdoses from reaching an all-time high.3 A great imbalance in our national approach to public safety, one that relies too heavily on the criminal legal system, has produced excessive levels of punishment and a diversion of resources from investments that would strengthen the capacity of families and communities to address the circumstances that contribute to crime. This report offers five recommendations for policymakers and community members to potentially improve safety without deepening our reliance on extreme sentencing:

Washington, DC: The Sentencing Project, 2023. 10p.

By Christopher Lowenkamp, Matthew DeMichele, and Lauren Klein Warren

This report presents findings related to impacts associated with criminal justice improvements underway in Lucas County, Ohio. The current report, however, focuses on impacts related to one of Lucas County’s initiatives - the use of the Public Safety Assessment (PSA). The report shows that Lucas County made serious reductions to the number of people booked into jail during the post-PSA period. For the entire seven-year study period, of those released pretrial, 27% experienced a failure to appear (FTA), 17% were arrested for any offense, and 5% were arrested for a violent offense during the pretrial period. There were reductions in the pretrial outcomes between the pre- and post-PSA periods: a six-percentage point decrease in FTA rates (30% vs. 24%), a five-percentage point decrease in new criminal arrest (NCA) rates (20% vs. 15%) and a two-percentage point decrease in new violent criminal arrest (NVCA) rates (6% vs. 4%). The results demonstrate that the PSA meets standards of predictive validity. For the three scales, we found that the Area Under the Curve (AUC) values are in the good (NCA) and fair (NVCA and FTA) ranges, there is incremental increase in failures as scores increase, and significant increases in the predicted likelihood of failure as scores increase across a series of regression models. The report shows that the PSA meets validity standards used for criminal justice assessments, and the report includes tests for predictive bias.

Advancing Pretrial Policy and Research, 2020. 84p.