Benchmarks and Evaluations - Safety versus Capabilities in Machine Ethics

Watch a comprehensive lecture from UC Berkeley's EECS series featuring Bo Li, Associate Professor from the University of Chicago, as she delves into the critical aspects of large language models, focusing on benchmarking methodologies, evaluation techniques, and the delicate balance between safety considerations and model capabilities. Explore the complex landscape of machine ethics while gaining insights into how these factors influence the development and deployment of AI systems. Learn about the various frameworks used to assess model performance, understand the inherent tensions between enhancing AI capabilities and ensuring safety protocols, and examine the ethical implications that arise in the field of artificial intelligence.