Mark_Friedenbach comments on Some alternatives to “Friendly AI”

[Mark_Friedenbach]

I’ll say again, “high assurance AI” better captures everything you described than “AI safety”.