I guess he was talking about the kind of precision more specific to AI, which goes like "compared to the superintelligence space, the Friendly AI space is a tiny dot.

But that stance makes assumptions that he does not share, as he does not believe that AGI will become uncontrollable.