'An AI hallucination nearly sparked an attack on a Chinese ship': X users debate slowing AI
binarybits · x · 2026-09-21
A debate over whether AI development should be slowed:
- binarybits finds it reasonable to slow AI because it may be destabilizing in hard-to-predict ways, comparable to the internet or the internal combustion engine.
- davidshor pushes back hard: the world is fragile, and he claims an AI hallucination nearly led the US to launch an assault on a Chinese ship last week — evidence AI could already kill many people, even short of existential risk.
The crux: is AI risk a gradual systemic disruption, or is it already triggering real geopolitical near-misses?
Related event: Debate Reignites Over Whether to Slow AI Development(2 posts)→
More from AGI Musings
- Uncle Bob on AI risk: install guardrails but push as hard and fast as we can — mark_k · 2026-09-21
- Molecule team to discuss agentic AI reshaping healthcare value chain at Roche's RoX Health event — melnykowycz · 2026-09-21
- Dartmouth Provost's Op-eds Flagged 96% AI-Written by Detector With 'Near-Zero Error' Audit — paulnovosad · 2026-09-21
- What if your gut feeling is just as conscious as you are? A thought experiment on the unified self — moultano · 2026-09-21
- Gary Marcus: the whole field joined the hybrid/neurosymbolic AI revolution and nobody admits it — GaryMarcus · 2026-09-21
- The Zvi on why AI risk isn't solvable: "Someone else builds Skynet" — TheZvi · 2026-09-21