AI bias

Abstract

Abstract This chapter examines how artificial intelligence (AI) bias can undermine legal (or legally relevant) norms and standards. It does so by introducing a conceptual distinction between bias in AI (arising from flawed data, programming choices, or emergent algorithmic behaviour) and bias towards AI (where human decision-makers either overtrust or unjustifiably dismiss AI outputs). This distinction can equip legal practitioners with a deeper, yet straightforward understanding of various AI biases and the risks they raise. To mitigate these risks, the chapter explores preventive and corrective strategies, including regulatory sandboxes, fairness-aware AI design, auditing laws, and legal oversight mechanisms. Addressing AI bias is not merely a technical challenge—it is a professional responsibility for legal practitioners who seek to properly navigate the relationship between law and AI.

Authors 1

1. Eduard Fosch‐Villaronga Aachen

   Leiden University

   Affiliation as printed

   Leiden University