The research field trying to explain, from a model's internal structure, why it produced a specific output — instead of only checking whether that output happens to be right.
Continue to AI University →