Algorithmic bias occurs when AI and machine learning models produce unfair or discriminatory outcomes due to skewed training data or flawed design. Addressing bias requires detection techniques, diverse data, and governance practices to ensure fairness.