The cosmic identity and related issues have been considered and I even used them to make a conjecture about alignment. As for role-playing games, I doubt that they are actually useful. Unless, of course, you mean something like Cannell’s proposal.
As for “the idea of arms races and the treacherous turn”, the AI-2027 team isn’t worried about such a risk, they are more worried about the race itself causing the humans to do worse safety checks.
The cosmic identity and related issues have been considered and I even used them to make a conjecture about alignment. As for role-playing games, I doubt that they are actually useful. Unless, of course, you mean something like Cannell’s proposal.
As for “the idea of arms races and the treacherous turn”, the AI-2027 team isn’t worried about such a risk, they are more worried about the race itself causing the humans to do worse safety checks.
But slightly irrational actors might not race (especially if they know that other actors are slightly irrational in the same or compatible way.)