Not only that, 3b seems to presuppose that the only dangerous AI is a recursively self-improving one.
Not only that, 3b seems to presuppose that the only dangerous AI is a recursively self-improving one.