In this interview he is suggesting that people stop a self-fufilling prophecy before it is to late. I think his fear is that if people believe superior ai is in

In this interview he is suggesting that people stop a self-fufilling prophecy before it is to late. I think his fear is that if people believe superior ai is inevitable that researchers will work off of those assumptions as if they were given facts - eventually creating systems which might pose threats or general harms to humanity. Other than that, only time will tell who is more correct.