Don't Worry About the Vase Podcast AI #155: Welcome to Recursive Self-Improvement
6 snips
Feb 12, 2026 Collection of quoted commentators including Eliezer Yudkowsky and others weigh in. Brief bios: Yudkowsky (AI researcher/thinker) and various commentators provide rapid-fire takes. They discuss fast frontier model upgrades, debates over mundane utility, agents and memory, safety and alignment worries, job displacement, jailbreaks, and political and governance tensions. Short, punchy perspectives throughout.
AI Snips
Chapters
Books
Transcript
Episode notes
Super Bowl Ads Missed The Public
- Super Bowl ads revealed two classes: frontier model branding and misleading productivity claims.
- Zvi argues many ads misrepresented capabilities and failed to educate mainstream viewers.
Jobs Will Be Radically Reshaped
- AI will replace many current jobs; using AI better only delays displacement.
- Zvi sees continuous reinvention as an unstable long-term survival strategy.
Experts Judge By Survival, Not Polish
- Experts measure artifacts by survival under pressure; outsiders judge polished artifacts.
- Zvi highlights that AI can do many subtasks, enabling non-experts to substitute for experts incrementally.
