Year: 2025
Type: conference-paper
Abstract: The use of algorithmic models or systems in criminal justice is increasing day by day, yet the bias in these algorithms can perpetuate historical inequities, especially in predictive tools like COMPAS. This literature survey examines 30 studies addressing algorithmic bias in criminal justice. Key topics include bias types, bias detection metrics or variables such as demographic parity and equalized odds, and bias mitigation techniques like re-weighting and adversarial debiasing. Challenges in achieving fair and unbiased predictions are highlighted, including ethical considerations and trade-offs or a balance between fairness and accuracy. Insights from COMPAS and similar systems underscore the need for ongoing research, proposing potential directions for policy and practice.
Institutions: Lovely Professional University, Nusa Putra University
Language: English
Subfield: Safety Research
Field: Social Sciences
Domain: Social Sciences
Open Access status: gold
OpenAlex ID: W4414099787