Debate: You Can't 'Premove' Against Superintelligence, Precautions Must Come Before It Arrives
ctjlewis · x · 2026-09-21
ctjlewis and RokoMijic clash over AI-safety timing. RokoMijic argues the argument fails because "superintelligence" lacks a clear definition. ctjlewis clarifies he isn't claiming SI would reach back in time, but that you cannot "premove" against SI like in chess — precautions must be taken in advance, because once SI arrives it's too late.
More from AGI Musings
- Lawyer's take: the AI boom will create more legal work, not less — jkubicki · 2026-09-21
- Models will stall near human-level since pretraining data holds nothing smarter, researcher argues — burny_tech · 2026-09-21
- FutureHouse publishes 'Millennium Problems for Biology' as wetlab-verifiable evals for AI — burny_tech · 2026-09-21
- Heidy Khlaaf clashes over 'rogue agent' narrative: bad cybersecurity and reward hacking can coexist — burny_tech · 2026-09-21
- Figma designer pushes back on Eric Schmidt's claim that UI will largely disappear — _AustinCalvert_ · 2026-09-21
- Vercel CEO: 'Fewer lawyers' is the dumbest AI prediction — jkubicki · 2026-09-21