Your Undivided Attention Ask Us Anything 2025
104 snips
Oct 23, 2025 In a whirlwind of AI developments, frontier labs are rushing to create smarter and faster models, raising questions about the future of work and human relationships with technology. Listeners explore the motivations behind tech companies' relentless push for new AI, even at societal costs. The complexity of AI's evolution comes under scrutiny, questioning if AGI might already exist. Insightful discussions also advocate for a benevolent AI future, emphasizing accountability in recruitment practices and the need for new governance structures to handle these rapid changes.
AI Snips
Chapters
Books
Transcript
Episode notes
Scaling Compute Produces Superhuman Skills
- Superhuman performance already exists in narrow domains via search and compute scaling.
- Continued compute and energy scaling make higher-than-human performance plausible across more tasks.
Use Collective Choice To Limit Risky AI
- Treat AGI as a collective choice and consider saying no until safe consensus exists.
- Build processes to decide which powerful systems we should or should not develop.
Incentives Define AI Outcomes
- A good AI future depends on aligning incentives, not just technology.
- Changing incentive structures is the core task to avoid negative externalities from AI.
