Algorithmic bias

Algorithmic bias

Systematic and unfair skew in an AI system's outputs, typically arising because the training data reflects existing social biases, or because of choices made in how a model is designed, evaluated, or deployed. Algorithmic bias has real-world consequences in areas such as hiring, credit scoring, and predictive policing, where flawed systems can reproduce or amplify existing discrimination at scale. (See also: Training data, AI alignment)