Book Title: Computational Criminology: AI Applications in Forensic Science and Criminal Justice
Editors: Dr. Xavier Louis, Dr. Surbhi Girdhar, Ms. Aswathi Chandran Nair, Mr. Ravi Kumar, and Ms. Nandini Katare
Chapter: 27
DOI: https://doi.org/10.59646/704/27
Author: M. Oviya
Abstract
The deployment of complex machine learning models neural networks, gradient boosting ensembles, deep learning architectures in criminal justice decision-making raises a fundamental challenge to the procedural values of the legal system: these models generate predictions whose internal logic is opaque, even to their creators. The right to understand the reasons for adverse decisions affecting one’s liberty is foundational to due process; the opacity of black-box algorithms renders these right formally hollow when algorithmic outputs drive criminal justice decisions. This chapter examines the technical landscape of explainable AI (XAI) methods including LIME, SHAP, attention visualisation, and inherently interpretable model architectures and evaluates their adequacy for the specific requirements of forensic and criminal justice contexts. It analyses the legal right to explanation under various constitutional and statutory frameworks, surveys judicial and regulatory responses to algorithmic opacity, and argues that genuine interpretability in criminal justice AI requires not just technical explanation tools but a fundamental commitment to models whose decision logic is comprehensible to affected individuals, their advocates, and adjudicators.