In Michigan, incarcerated men engage in employment counseling within state prisons. Their access to job training and educational opportunities varies based on security classifications, which are now under scrutiny. Michigan is testing a new system aimed at improving predictions of potential prison violence.
State officials claim that the new system could enhance predictions of future prison violence by 30%. The accuracy of these predictions is essential for the safety of correctional officers and inmates.
The Role of AI in Michigan Prisons
Initial reports suggested artificial intelligence might play a role in evaluating security risks. However, a nonprofit organization developing the software clarifies that AI is not part of calculating security scores. Instead, the system adheres to set rules, with corrections staff making final decisions.
The system will undergo testing starting December 1 in select men’s prisons. If successful, it may extend to more facilities. Currently, corrections staff reassess a person’s security level at scheduled intervals. The new system allows for ongoing reviews with updated prison records, potentially identifying behavioral changes sooner.
Scoring System Mechanics
Developed by Recidiviz, a technology nonprofit serving criminal justice agencies in 19 states, the system uses analyzed data from the past decade of Michigan prison classifications. It includes recent disciplinary incidents and good behavior as criteria for scoring, aiming to align security levels with current behaviors and needs.
Qualification for lower security levels can be influenced by completed rehabilitation programs. The system allows staff to review calculations and adjust scores as necessary. This ensures that medical needs or personal safety concerns can factor into the final security classification.
Challenges Beyond Technology
Despite potential improvements in classification accuracy, Michigan faces other challenges, such as a shortage of higher-security beds. Recidiviz’s point-based formula must sometimes contend with the unavailability of appropriate housing, leading to concerns raised by corrections staff unions.
The State’s Safe Prisons Initiative aims to add 870 high-security beds, with ongoing facility conversions and new construction projects expected to take over two years. However, timely solutions are necessary to address current overcrowding issues.
Conclusion: Balancing Technology and Human Judgment
The new scoring system in Michigan presents an opportunity to enhance safety and fairness within state prisons. Yet, technology alone cannot resolve issues of space and staffing shortages. Moreover, understanding how the expected 30% improvement in violence prediction was calculated remains crucial.
Decisions impacting inmates’ lives should involve careful analysis and human judgment, especially given the potential for errors in the underlying data. Michigan’s pilot program may signal progress, but results will ultimately determine the system’s effectiveness and the balance between automated and human decision-making.