An AI system used in parole decisions is challenged in court because it cannot provide the basis for individual recommendations. Which AI ethics principle is MOST at stake?

The correct principle is explicability/transparency, because the court challenge centers on the system's inability to provide meaningful explanations for individual parole recommendations. In consequential decisions affecting individual liberty, individuals have a right to understand the reasons for those decisions, and in criminal justice contexts this also raises a due process concern. The fairness option may be relevant if racially biased recommendations were alleged, but the challenge described specifically concerns the inability to explain individual decisions. The privacy option concerns sensitive criminal justice data, but that is not the focus of the described challenge. The beneficence option concerns maximizing public safety outcomes, but the issue raised is the lack of explanation for individual recommendations.