x
Demons in the Shell — Why Current A.I. Alignment Methods are Creating the Very Monsters We Should Fear — LessWrong