Ezra Klein's Podcast with Jensen Huang
4 hours ago
- Jensen Huang does not believe in superintelligence or AI existential risk, viewing AI as just another software abstraction layer.
- He insists on engineering rigor and product safety, claiming labs should not ship unsafe products and should spend 10x compute on verification.
- He inadvertently argues for shutting down AI labs if they cannot contain experiments, contradicting his opposition to slowing down.
- He denies that competitive pressure forces labs to accelerate, despite clear evidence, and resists antitrust exemptions for safety collaboration.
- He confuses AI safety with ordinary product safety, failing to grasp unique challenges like alignment and emergent behavior.
- His positions as Nvidia CEO significantly influence AI policy and chip allocation, making his misunderstanding dangerous.