>> "A machine intelligence need not share any of that complexity"

I think this is an assumption that both Bostrom and yourself would acknowledge as such.

We simply don't know enough about the nuances of intelligence to make an assertion that there are inherently varying kinds of it.