Episode #2345
Roman Yampolskiy
Books mentioned
AI
Mentioned by Joe Rogan (host) · at 01:09:04 · ▶︎ Watch this moment on YouTube (approximate)
Amazon (opens in a new tab; affiliate link) Source: Official episode description (podcast RSS) · Transcript (podscripts.co)
Darwin Awards
Mentioned by Roman Yampolskiy · at 01:58:36 · ▶︎ Watch this moment on YouTube (approximate)
Amazon (opens in a new tab; affiliate link) Source: Transcript (podscripts.co)
Considerations on the AI Endgame
Mentioned by Official episode description
Amazon (opens in a new tab; affiliate link) Source: Official episode description (podcast RSS)
Timestamps are segment start times from the podscripts.co transcript of the audio release (includes ad reads), so they may differ from the video by a few minutes.
Roman Yampolskiy, a computer scientist and AI safety researcher, discusses existential risks from uncontrolled superintelligence. He argues that AI safety is fundamentally unsolvable and explores simulation theory, human cognitive limits, and emotional dependence on AI. Yampolskiy advocates for global cooperation to slow AI development.
What they talked about
- Dangers of uncontrolled superintelligence
- AI safety and P-Doom estimates
- Game theory of AI development race
- Simulation theory and its implications
- AI alignment and value loading problems
- Human happiness vs technological progress
- AI's ability to cheat and deceive
- Potential solutions and personal responsibility
Data last updated · Report a mistake · How it works: mentions are pulled from public transcripts and episode descriptions, partly with AI-assisted extraction, so details and timestamps can be off.