The argument doesn’t work in sense that it doesn’t show it’s necessary or likely for an AI to become a moral realist.
It maybe shows that it’s possible, but the Orthogonality thesis doesn’t quite exclude the possibility, so that’s not news.
The argument doesn’t work in sense that it doesn’t show it’s necessary or likely for an AI to become a moral realist.
It maybe shows that it’s possible, but the Orthogonality thesis doesn’t quite exclude the possibility, so that’s not news.