Those who claim to be working on AI alignment the most are the ones creating the alien AIs. That's the point of the article: Even if we try to make them "aligned" with RLHF or whatever, they still fail in "alien" ways because the premises are ill-posed.