Where this was said
Humanity in the Cockpit: Governing the Ungovernable
At 1:59:00 · chapter starts 1:56:40
Bostrom synthesizes the governance problem with a metaphor that immediately lands: the pilot has had a heart attack. The passengers have to figure out how to fly the plane. Add to that — all the passengers are in a fistfight in the cockpit, each convinced they are the superior person to land it. That is the current state of AI governance: humanity trying to shepherd civilization's most dangerous and consequential technology toward a good outcome while squabbling monkey tribes compete for control of the controls. Rogan offers his own prediction: the first domain AI will visibly take over is government, because the inefficiency and fraud are so extreme that a genuinely intelligent system will find them obviously fixable. Bostrom gently notes that some people might not want that particular application. The two discuss the possibility that once a sufficiently good AI exists, humans can hand over the reins — and might only need to get things 'so far' before help arrives.
The pilot has passed out. The passengers have to fly the plane. But they're also in a fistfight over who should be at the controls. That's the current state of AI governance — humanity trying to land civilization's most dangerous technology while being a squabbling monkey tribe.
