"The Cognitive Revolution" | AI Builders, Researchers, and Live Player Analysis Agency over AI? Allan Dafoe on Technological Determinism & DeepMind's Safety Plans, from 80000 Hours
203 snips
Mar 15, 2025 
Allan Dafoe, Director of Frontier Safety and Governance at Google DeepMind, discusses the vital intersection of AI safety and governance. He explores technological determinism versus human agency, showcasing how societal choices shape technological outcomes. Dafoe delves into the ethical responsibilities in AI development, emphasizing collaboration to avert an AI cold war. He also examines structural risks and highlights the need for proactive safety protocols in AI alignment. The conversation sheds light on the transformative potential of AI in various sectors.
AI Snips
Chapters
Transcript
Episode notes
Nuclear Crew Anecdote
- Rob Wiblin's uncle, part of a nuclear weapons crew, planned to go AWOL if ordered to launch.
- This anecdote highlights individual agency even in high-stakes, seemingly deterministic situations.
Situational Awareness
- Frontier AI developers should increase situational awareness, even if it slows project progress.
- Ensure the overall enterprise aligns with personal values.
Ethical Considerations
- AI developers should consider which project directions might lead them to abandon the project.
- This proactive reflection prepares for potential ethical conflicts in the evolving AI landscape.