My big concern with misdiagnosing autism in high-IQ children is that it conflates autism with introversion -- a conflation that has been progressing for some time now, and which has been accelerating in recent decades. This is bad for the legitimately autistic, and it's bad for introverts.
It's bad for the autistic because it lumps a fairly heterogeneous set of children into the same spectrum, confounding our best efforts to get to the true roots of autism. If we're treating too many nonautistic children as "autistic," then by nature, we're diluting the actual pool of people we should be studying and helping. We're throwing too many red herrings and exceptions into the data set.
It's bad for the introverted and highly intelligent because it further stigmatizes both of those things, especially in combination. By establishing extroversion and median IQ as "ideal," we're insinuating that the introverted and gifted are somehow defective. We're behaving as though our gifted children should be bent and twisted into acting more like everyone else, when in fact, we should be nurturing their gifts.
While autism-spectrum disorders, introversion, and high IQ can certainly overlap, they are not by nature the same things.