This chapter analyzes bias detection and fairness assessment in machine learning systems, a hot topic in artificial intelligence. We start by listing all sorts of bias in the machine learning pipeline: historical, representational, measurement, aggregation, and evaluative. We will next build a rigorous mathematical foundation for fairness and specifically define essential notions like statistical parity, equated chances, equal opportunity, and calibration, demonstrating that many fairness principles cannot be achieved concurrently. We then construct effective bias detection methods employing exploratory data analysis, model assessment, and statistical testing. We apply frameworks to measure fairness across stakeholders and delivery conditions. Pre-model data debiasing, fairness constraints in the learning algorithm (in-processing), and model output adjustments (post-processing) are examined to reduce bias. We use compelling loan and healthcare case studies to demonstrate bias detection and mitigation systems’ practical use and the complex trade-offs between fairness and accuracy. Finally, we explore advancing prejudice and fairness studies. We briefly cover causal fairness, long-term fairness dynamics, and basic model fairness, among other exciting new study areas. .