Data bias, intelligent systems and criminal justice outcomes.

This article examines the increasing use of artificial intelligence systems (AIS) in the criminal justice system and focuses on the pervasive problem of data bias affecting AI-driven outcomes. Data bias—manifesting in various forms such as historical, sample, representation, and measurement bias—can lead to unfair and discriminatory justice outcomes, including breaches of human rights like equality, fair trial, privacy, and freedom of expression. The article highlights AI applications in courts, law enforcement, and corrections, such as predictive tools for recidivism and facial recognition, noting their benefits alongside risks of bias and opacity. To address these challenges, it advocates a multifaceted approach combining legal reforms (e.g., leveraging provisions like Article 22 of the EU General Data Protection Regulation), establishment of skilled regulatory bodies within meta-regulatory frameworks, specialized diversity and bias training for stakeholders, and adherence to ethical principles of fairness, accountability, transparency, and explicability in AI design and deployment.