teaching

Frameworks for evaluating and understanding AI systems. Covers measurement as predictive modeling, measurement validity, and benchmark design and governance.

Building ML systems that learn from human preference data. Topics include supervised learning from preferences, active elicitation strategies, and reinforcement learning approaches.

Methods for capturing human and societal preferences in AI systems. Covers inverse reinforcement learning, metric elicitation, and reinforcement learning from human feedback.

Mechanisms for capturing human feedback to design reward functions. Covers inverse reinforcement learning, metric elicitation, and reinforcement learning from human feedback.