A Sociotechnical Approach to a Safe, Responsible AI Future: A Path for Science and Evidence-based AI Alignment and Trust

A Sociotechnical Approach to a Safe, Responsible AI Future: A Path for Science and Evidence-based AI Alignment and Trust

Watch a 24-minute lecture from UC Berkeley professor Dawn Song at the Simons Institute exploring critical aspects of responsible AI development, focusing on alignment, trust mechanisms, watermarking techniques, and copyright challenges in Large Language Models (LLMs). Gain insights into the sociotechnical framework needed to ensure safe AI advancement through science and evidence-based approaches. Learn about current challenges and proposed solutions in creating trustworthy AI systems while understanding the complex interplay between technical implementation and social responsibility.