AI-Powered Resentencing: Bridging California’s Second-Chance Gap Through Natural Language and Vector-Based Legal Analytics

• In Event: Lightning Session: Crime, Justice, and Drugs & Employment and Training Programs

Abstract

Resentencing presents a complex legal challenge despite groundbreaking legislative reforms like California's Racial Justice Act (2020) and Prosecutor Initiated Resentencing (2018). Implementation lags behind legislative intent, creating a 'second-chance gap' where hundreds of eligible cases remain unidentified within overwhelming caseloads. This paper demonstrates how interpretable AI systems and analytics capabilities can democratize legal research and accelerate case evaluation through natural language interfaces and vector-based searches. Our open-source platform processes approximately 95,000 prison records, enabling attorneys to systematically identify similarly situated defendants, identify racial biases in sentencing, quantify rehabilitation progress, and generate court-ready evidence for prima facie and discovery motions. We present mathematical frameworks for case similarity scoring and suitability assessment, alongside real-world deployment results showing over 450 potential cases identified for review across California counties. This work illustrates how ethical data science can augment legal expertise to address systemic injustices while maintaining transparency and interpretability.

Author

• Aparna Preeti Komarla, Independent
  Presenting Author