tszzl: AI should be treated as more dangerous than nukes, alignment is Plan A
tszzl · x · 2026-11-04
Researcher tszzl lays out his big-picture AGI safety view: the ideal is international agreements treating AI as more dangerous than nuclear weapons, but he doesn't expect global mutually-assured coordination to arrive quickly without destructive demonstrations. Since no plan can stop everyone from building this, solving sufficient alignment is "Plan A." He argues bad outcomes are inevitable — someone will hook AI to the internet, hacker groups will weaponize it, non-alignment groups will lose control of their own systems, and rogue AI will self-improve — so reasoning about a world where everything goes to plan is pointless.
More from AGI Musings
- US datacenter buildout stalls while Paraguay and Kazakhstan scoop up AI compute — ivan_bezdomny · 2026-11-04
- AI is a disposable resource, not a moat: use it to build non-AI advantages — paraschopra · 2026-11-04
- Baseten engineer argues compute and experiment bottlenecks make 2027 RSI, 2028 superintelligence forecasts too optimistic — AccBalanced · 2026-11-04
- Ramez Naam: power concentration is the biggest AI risk, plural AI makes us safer — sebkrier · 2026-11-04
- Sequoia summit: the diffusion gap between model progress and enterprise adoption is the opportunity — FinanceYF5 · 2026-11-04
- Sequoia summit: ease of use is underrated; build around customers, not model releases — FinanceYF5 · 2026-11-04