Must be in the Computer Science Master's or Doctoral program; or permission of the instructor. Assumes knowledge of natural language processing, computation graphs, and Python proficiency.

Getting to computers to answer questions is arguably the foundation of AI (the Turing Test). The course covers the historical approaches to question answering with computational approaches, the philosophical and theoretical underpinnings of how and why questions are asked, and how tocreate effective questions and reliable answers. The course will then move to evaluation: how to accurately rank systems (leaderboards) on notjust accuracy but calibration and what happens when you have multiple potential correct answers to a question (pluralistic alignment).