Explore a comprehensive lecture on Computational Social Choice and AI Value Learning delivered by Rachel Freedman at the 2024 Cooperative AI Summer School. Delve into the intersection of social choice theory and artificial intelligence as Freedman, a PhD student at the Center for Human-Compatible AI at UC Berkeley, shares her expertise. Gain insights into misspecification problems in Reinforcement Learning from Human Feedback (RLHF), model interpretability and control, and dangerous capabilities evaluations for foundation models. Over the course of 55 minutes, discover how these concepts contribute to the development of AI systems that align with human values and societal preferences. Enhance your understanding of the challenges and opportunities in creating AI systems that can effectively learn and incorporate human values in decision-making processes.
Overview
Syllabus
Computational Social Choice and AI Value Learning by Rachel Freedman
Taught by
Cooperative AI Foundation
Related Courses
-
Reinforcement Learning from Human Feedback (RLHF)
-
Generative AI Advance Fine-Tuning for LLMs
-
RLHF: How to Learn from Human Feedback with Reinforcement Learning
-
LLM Alignment - Techniques for Building Human-Aligned AI
-
Aligning AI to Everyone via Reinforcement Learning
-
Reinforcement Learning from Human Feedback (RLHF) Explained
