I think that's a fair way of putting alignment. The concern is that we might make an AI that's capable of outsmarting us before we're good at making it care abo

I think that's a fair way of putting alignment. The concern is that we might make an AI that's capable of outsmarting us before we're good at making it care about doing what we want instead of some other goal that disregards what we really care about like our safety.