Dan Hendrycks joins the podcast again to discuss X.ai, how AI risk thinking has evolved, malicious use of AI, AI race dynamics between companies and between militaries, making AI organizations safer, and how representation engineering could help us understand AI traits like deception.
Dan Hendrycks joins the podcast again to discuss X.ai, how AI risk thinking has evolved, malicious use of AI, AI race dynamics between companies and between militaries, making AI organizations safer, and how representation engineering could help us understand AI traits like deception. You can learn more about Dan's work at https://www.safe.ai Timestamps: 00:00 X.ai - Elon Musk's new AI venture 02:41 How AI risk thinking has evolved 12:58 AI bioengeneering 19:16 AI agents 24:55 Preventing autocracy 34:11 AI race - corporations and militaries 48:04 Bulletproofing AI organizations 1:07:51 Open-source models 1:15:35 Dan's textbook on AI safety 1:22:58 Rogue AI 1:28:09 LLMs and value specification 1:33:14 AI goal drift 1:41:10 Power-seeking AI 1:52:07 AI deception 1:57:53 Representation engineering
Maya Ackerman discusses human and machine creativity, exploring its definition, how AI alignment impacts it, and the role of hallucination. The conversation also covers strategies for human-AI collaboration.
Adam Gleave, CEO of FAR.AI, discusses post-AGI scenarios, risks of gradual disempowerment, defense-in-depth safety strategies, scalable oversight for AI deception, and the challenges of interpretability, as well as FAR.AI's integrated research and policy work.
Beatrice Erkers discusses the AI pathways project, focusing on approaches to maintain human oversight and control over AI, including tool AI and decentralized development, and examines trade-offs and strategies for safer AI futures.
