> especially the idea that AI itself is an agent in defining its own future

Why? I see no evidence that this won’t be the case.. or isn’t already