turchin comments on The virtual AI within its virtual world

[turchin]

For me Safe AI is one that is not existential risk. “Friendly” reminds me about “friendly user interface”, that is something superficial for core function.