As the lead inventor of this particular piece of bogus technology (Tom N. worked on the project under my direction) I'll politely disagree. Of course, popular articles oversimplify -- we don't expect to be able to identify the species of a single mosquito on the wing with 100% accuracy.
We can, however, measure the frequency at which an individual mosquito flaps its wings, and, on average, that differs from species to species, and differs quite a lot between males and females of a given species. So we can indeed tell individual males from females, and with wingbeat frequency and other data we expect to be able to get a pretty good statistical estimate of the population distribution among species in a given area. And indeed, we are very interested in applications of the system, without the killing mechanism, to collect data on insect populations.