The Importance of Proper Guardrails for AI Development

All the top labs in the world are really concerned at various levels about catastrophic risk from AI accidents and other things. As you zoom in on the safety people, the people who think about this problem the most, they tend to believe even more highly that this is a thing. And so it's hard to walk away and be like, okay, there's less than a 10% chance of this happening when so many serious people seem to be taking it so seriously.