I've studied AI risk for 20 years. We're close to a disaster.

Roman Yampolskiy explains why superintelligence cannot be controlled, why the gap between AI capabilities and AI safety keeps widening, and how narrow AI could deliver cures and productivity without risking extinction. 00:00 Rogue AI 00:30 Deception 03:28 Existential risk 06:10 Losing control 10:54