When an AI system's outputs are systematically less accurate or less favorable for people in a specific demographic group, in a way traceable to how the model was built or trained.
Continue to AI University →