lemire: 'superintelligence will kill us all' is a moral claim, not a rational one

lemire · x · 2026-09-10

Daniel Lemire argues statements like 'a superintelligence will kill us all' are moral claims, not rational ones—akin to 'gas cars destroy the planet.' They have a rational basis, but in practice the rational basis is secondary: proponents need not provide evidence, and counter-evidence won't move them, which is why calls for rational explanations from labs go unanswered. He cites Robin Hanson's complementary take on the endless emergence of vague but terrible fears. The quoted cmuratori challenge sharpens it: if Anthropic must build ASI to stop OpenAI, what exactly would its first human-extinction-capable ASI do to prevent competitors from catching up? That first-mover logic has never been spelled out publicly.

Original post →

More from AGI Musings

AGI Musings channel →