I read GP trying to draw a distinction between “real ethics of AI in practice” and “fantasies about scary overlord general AI.”

The former sounds like it would be full of reasonable cases for taking it seriously.