Jensen Huang vs. the A.I. Doomers
About this episode
Huang, Nvidia's CEO, presents AI as a five-layer 'cake' — energy and chips, AI factories, models, and the applications layer he cares about most — and pushes back hard against AI alarmism. Distinguishing a job's purpose from its tasks, he argues automation (as with radiology) expands demand rather than destroying work, citing human ambition and $500 billion in recent AI venture investment as proof of net job creation. He defends open-weight models (Nvidia just bought Hugging Face for about $12 billion), deflates the OpenAI agent-hacking incident as a solvable engineering failure of sandboxing and alignment, and resists new regulation: existing liability law plus corporate responsibility suffice — 'don't ship' unsafe products. He calls Geoffrey Hinton's doom predictions irresponsible and unscientific, rejects the 'phase change' framing of machine intelligence as anthropomorphized software, and treats recursive self-improvement as ordinary computer engineering.