I think we're now largely in agreement, about the destination at least. Whether there is a stage of human augmented AI or not is moot (but I'd agree with @drehb's quote above, which makes me think it's less likely than more).
I don't know how many millions of years it took to go from the first primates to modern humans, but once our brain reached the threshold to move from genetic "learning" (ie evolution by natural, genetic processes) to cultural learning/evolution the subsequent leaps have happened so fast there has been hardly any generic component to them.
In about 100,000 years we have jumped, in ever reducing timescales, from one technology (human augmented intelligence level if you like) to the next: to make stone tools, fire, language, metals, writing, printed books, science, electricity, radio, rockets, computers, atomic bombs and each revolution quicker than the last.
So it's logical to think human augmentation will continue. But why, when intelligence reaches the stage where it can design the host itself? Humans are just beginning to be able to do that (eg creating new life by designing a genome), but I think AI will zip past that stage very quickly and have little use for the human body in the process.
In my life I've watched these kind of innovations compress in time from taking decades to, about a decade, now faster than that and still accelerating. Remove humans and our slow learning processes from the loop and bang!
It's just my opinion, but I think that things will evolve so fast that if we blink we'll miss any super intelligence stage that is recognisably human or based on the human genome.
I find it hard to believe that this body of mine is, by chance, a component in the best solution to the next big leap in intelligence / evolution.