Future of Life Institute Podcast

Dan Faggella, AI expert and entrepreneur, discusses AGI implications, AI power dynamics, industry implementations, and what drives AI progress in a thought-provoking podcast conversation.

Liron Shapira joins the podcast to discuss superintelligence goals, what makes AI different from other technologies, risks from centralizing power, and whether AI can defend us from AI. Timestamps: 00:00 Intelligence as optimization-power 05:18 Will LLMs imitate human values? 07:15 Why would AI develop dangerous goals? 09:55 Goal-completeness 12:53 Alignment to which values? 22:12 Is AI just another technology? 31:20 What is FOOM? 38:59 Risks from centralized power 49:18 Can AI defend us against AI? 56:28 An Apollo program for AI safety 01:04:49 Do we only have one chance? 01:07:34 Are we living in a crucial time? 01:16:52 Would superintelligence be fragile? 01:21:42 Would human-inspired AI be safe?

Annie Jacobsen, an expert on nuclear war, lays out a second-by-second timeline for nuclear war scenarios. Discussions include time pressure, detecting nuclear attacks, decisions under pressure, submarines, interceptor missiles, cyberattacks, and concentration of power.

Katja Grace discusses AI researchers' beliefs, discontinuous progress, impacts of AI crossing human-level intelligence, intelligence explosions, and mitigating AI risk in the largest survey of AI researchers. Topics include AI arms races, slowing down AI development, and intelligence and power dynamics. Grace explores high hopes and dire concerns, AI scaling, and what AI learns from human culture.

Discussion on pausing frontier AI, risks during a pause, hardware overhang, safety research, social dynamics of AI risk, and the challenges of cooperation among AGI corporations. Also, explores the impact on China and protesting AGI companies.

Sneha Revanur, AI researcher, discusses the social effects of AI, ethics vs safety, humans in the loop, AI in social media, AIs identifying as AIs, AI influence in elections, and AIs interacting with human systems.

Roman Yampolskiy, AI researcher and expert, discusses whether AI is like a Shoggoth, scaling laws, evidence for AI being uncontrollable, and the safety of designing human-like AI. They also explore the limitations of AI explainability, verifiability, and alignment. The conversation touches on the challenges of integrating AI into society, deleting dangerous information from neural networks, building human-like AI for robotics, and potential obstacles to implementing language models in various industries. They conclude by discussing Good Heart's Law, a positive vision for AI, and the challenges of regulating AI investment.

Flo Crivello, an expert in AI and its implications for society, discusses AI as a new form of life, regulatory capture risks, the possibility of a GPU kill switch, and predicts AGI within 2-8 years. They also explore Biden's executive order on AI, regulating models or applications, and the collaboration between China and the US on AI. The podcast delves into the challenges of managing AI systems and the philosophical question of subjective experience in AI.

Carl Robichaud, an expert on nuclear arms race and nuclear risk, discusses topics such as the new nuclear arms race, the role of world leaders and ideology in nuclear risk, the impact of nuclear weapons on stable peace, North Korea's nuclear weapons, public perception of nuclear risk, and reaching a stable, low-risk era.

Frank Sauer discusses autonomy in weapon systems, killer drones, low-tech defenses against drones, flaws and unpredictability of autonomous weapon systems, and the political possibilities of regulating such systems.