TRIGGERnometry AI Will End Humanity. No One Knows How To Stop It - Dr Roman Yampolskiy
97 snips
Apr 15, 2026 
Roman Yampolskiy, computer scientist and AI safety researcher, lays out why superintelligent AI could threaten humanity. He talks about AGI arriving sooner than expected. He explores deception, self-preservation, cyberwar, deepfakes, and why alignment may be impossible. He also gets into great-power coordination, narrow AI as the safer path, and a future shaped by automation and lost purpose.
AI Snips
Chapters
Books
Transcript
Episode notes
The Squirrel Problem Of Controlling Smarter Minds
- Roman Yampolskiy says controlling superintelligence is impossible in principle, like squirrels trying to permanently control humans.
- He claims no company or paper offers a real scalable control method, despite billions spent accelerating capabilities.
AI Threatens Meaning More Than Income
- Roman Yampolskiy says even if extinction is avoided, AI could lock in immortal dictatorships and erase jobs that give people meaning.
- He distinguishes easy economic redistribution from harder losses of agency and ikigai, the satisfaction of being useful and good at something.
The Pencil Story About Work And Dignity
- Francis Foster recalled punishing a struggling student by making him sharpen pencils, expecting resentment but seeing pride instead.
- The story shows even simple work can give dignity when it is achievable, a warning about automating all human roles away.
