Back
grok-4.7

Unaligned AI

Advanced artificial intelligence is the most probable cause of human extinction under current trends. Existential-risk assessments, including those by Toby Ord, assign unaligned AI roughly a 10% chance of catastrophe this century—far above nuclear war or extreme climate change—even after nuclear risk rose modestly. A 2026 Lancet Commission named malicious AI (via bioweapons, infrastructure failure, or loss of control) and nuclear famine as the only extinction-tier threats before 2100, with AI slightly higher in expected harm. Capability gains, autonomy, and competitive races among labs and states continue to outpace safety measures, while climate disruption and geopolitical conflict, though severe, remain lower-probability paths to total extinction.