He totally misunderstands the danger of runaway AI's. Even perfectly intentioned AI researchers living in his imagined utopia would be at risk of accidentally creating one. Aligning AI's with human values is a non-trivial technical problem, and there's no guarantee we'll find a solution before someone creates a super intelligent AI, regardless of how well we structure society.