Graduate students register for PHIL 261C. In this seminar we look into what makes a good explanation through three complementary perspectives. Philosophy attempts to define the normative criteria that make a scientific theory explanatory. Cognitive science looks into the criteria implicit in our common sense choices of explanations. Explainable AI tries to operationalize these principles into protocols for rendering black-box models interpretable. We look into how the normative criteria of philosophers inform our cognitive theories, and how both provide perspective into the goals and methods of XAI.
4 units · Letter or Credit/No Credit
Graduate students register for 261C. In this seminar we look into what makes a good explanation through three complementary perspectives. Philosophy attempts to define the normative criteria that make a scientific theory explanatory. Cognitive science looks into the criteria implicit in our common sense choices of explanations. Explainable AI tries to operationalize these principles into protocols for rendering black-box models interpretable. We look into how the normative criteria of philosophers inform our cognitive theories, and how both provide perspective into the goals and methods of XAI.