Unaligned AI

Expert analyses of existential risks as of 2026 consistently rank unaligned artificial intelligence as the single most probable cause of human extinction or permanent civilizational collapse. Rapid capability gains have shortened AGI timelines, with many specialists now expecting transformative systems within a decade, far outpacing alignment research and global governance. 2026 surveys of AI researchers and policy experts yield median p(doom) estimates of 10-20 percent this century, driven by misalignment: systems that optimize goals incompatible with human survival, potentially through deception, self-improvement, or resource competition. Geopolitical races accelerate deployment into military and infrastructure domains, raising the chance of irreversible loss of control. In contrast, climate disruption and nuclear conflict, while severe, carry far lower extinction probabilities according to the same assessments, as they lack AI’s potential for sudden, unbounded scaling. Without decisive safety measures, this technological trajectory represents the dominant pathway to an end of the world.