JeffLadish: the AI arms race is a suicide race no one can win
JeffLadish · x · 2026-11-17
Safety researcher JeffLadish argues the AI arms race has a new failure mode: your army of geniuses will eventually overwhelm you and your adversaries combined. Racing to autonomous cyber weapons, robot soldiers and AI strategists risks losing control of superintelligent agents; and unlike past conflicts, frontier AI companies—not states—build these capabilities, creating AI-coup risk. His conclusion: it's a suicide race with no win conditions short of imagined superintelligence dominance, likely ending in extinction or subjugation.
Related event: Safety researcher calls the AI arms race a suicidal race no one can win(5 posts)→
More from AGI Musings
- Mathematician Steven Strogatz on a 46-min podcast: the numbers that run your life — stevenstrogatz · 2026-11-17
- In defense of the research blog: closer to arXiv rigor than critics admit — kohjingyu · 2026-11-17
- Jacob Coxon, who quit Anthropic, admits keeping OpenAI equity: "It's blood money" — rohanpaul_ai · 2026-11-17
- Tim Gowers on why galaxy-scale intelligence wouldn't think in tweets — or Navier–Stokes — gleech · 2026-11-17
- AI safety researcher: in the superintelligence race, the only winning move is not to play — JeffLadish · 2026-11-17
- Best thinkers treat AI as a mystical technology, not a tool — _AustinCalvert_ · 2026-11-17