Equility before the law and sentencing disparities
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
67 Equality before the law and sentencing disparities Abstract The thesis examines the phenomenon of recidivism through the lens of machine learning methods and compares them with a simple statistical model. The aim is to assess whether more complex algorithms (logistic regression, tree-based models, and a simple neural network) deliver a significant and defensible improvement in performance over a simple statistical model, and what regulatory requirements govern their deployment. An important finding is that even relatively simple machine learning methods yield substantial improvements over the baseline statistical model. The first part delineates the theoretical foundations of equality and the assessment of recidivism in criminal law. It further analyzes the new regulatory framework for artificial intelligence and its implications for the development, testing, and deployment of predictive models in criminal justice. This framework imposes a range of obligations (risk management, data governance, documentation, fundamental-rights impact assessment, human oversight) and simultaneously defines prohibited practices that must be considered in the design and operation of models. The thesis introduces modern methods and presents the best-known practical tool, COMPAS, including an explanation of its functioning...
Fields of Study & Subjects
Ethics and Social Impacts of AI
Artificial Intelligence in Law
Criminal Justice and Corrections Analysis
Publication details
- Journal / source
- Digital Repository (National Repository of Grey Literature)
- Publication date
- 1 January 2026
- Language
- CS
- Document type
- Thesis