I agree with your conclusion. Just a nitpick:

I wouldn't trust the strong version of the accelerating change hypothesis. Actually augment our intelligence, or at least our speed of thought probably require major breakthroughs in neuroscience and relevant fields in engineering, which may occur as not-so-predictable quantum leaps.

Plus, there the safety problem: if we ever enhance someone's intelligence, we may want to make sure that (i) it doesn't make him mad, and (ii) the guy doesn't plan to take over the world. The proper sanity checks may require yet other breakthroughs.

In the end, Friendly AI may be easier.