Caspar Oesterheld comments on Some alternatives to “Friendly AI”