Michigan Tests New Prisoner Classification System

Michigan will begin testing a new prisoner classification system on December 1, which officials say might improve predictions of future violence by about 30 percent, the Michigan Department of Corrections (MDOC) announced.

The pilot program comes in response to a notable increase in assaults within U.S. prisons. According to state figures, assaults on prison staff rose 77 percent, and assaults between incarcerated individuals increased 54 percent from 2019 to 2024.

The trial will commence in selected men’s facilities. Here, classifications can impact prisoners’ movement, supervision, and access to activities and programs.

Development and Implementation

The nonprofit technology organization Recidiviz helped develop the system. This classification tool, however, does not use artificial intelligence. Recidiviz confirmed that AI tools are not applied for decisions about a person’s security level or release. Michigan currently supervises about 32,000 incarcerated individuals and started revamping its classification approach due to the rising violence in prisons nationwide.

Last year, the state enlisted Recidiviz to assist in creating the new system. The goal was to evaluate its effectiveness before considering broader implementation.

Current and New Classification System

Michigan’s current classification system employs four security categories: Levels I, II, IV, and V. There is no Level III. These classifications significantly impact movement freedom, supervision levels, and activity access. Higher-security prisoners face stricter restrictions, while those at lower levels have broader opportunities, such as vocational training and higher education.

The new approach will incorporate real-time data analysis, rather than relying solely on static inputs collected at set intervals. According to MDOC, this should enhance predictions of future violence by approximately 30 percent, though the methods for calculating this improvement have not been disclosed.

Official Statements

Heidi E. Washington, MDOC Director, emphasized the system’s aim to assign security levels based on current behaviors and needs, aiming to protect both incarcerated people and corrections staff. “I am optimistic that this pilot’s use of evidence-based and historical data has created a tool that is dynamic, objective, and will strengthen the security of our facilities,” Washington noted.

MDOC intends to leverage this pilot to determine the tool’s effectiveness before a wider rollout.

Considerations and Concerns

Using algorithms for risk assessment in the criminal justice system has raised issues regarding transparency and potential bias. A ProPublica investigation into a different system used in Louisiana highlighted concerns about racial and socioeconomic disparities in data. While Michigan’s strategy focuses on prison security classifications, unlike the Louisiana system, similar questions about data sources, improvements in predictive accuracy, and possible bias monitoring remain unanswered.

Understanding the input data, verifying claimed accuracy improvements, assessing the possibility for staff overrides, and ensuring bias detection mechanisms are in place could be crucial as Michigan evaluates the permanence of this new classification system.