According to New Scientist, a comparison of smart fruit filed and average fruit flies in conditions where food was scarce, the smart ones died off quicker:
"They are slower at feeding," says Mery. He speculates that the flies may have to invest more energy in making or re-arranging connections between neurons in their brains, leaving them with less energy to forage when calories are limited. He cautions that the work measures the ability of fruit flies to learn only a particular task, though the group is testing whether their smart flies are also better at learning other things. "But this should open up the evolutionary question," he says. In principle, it should be possible to look for the costs of intelligence even in primates.
So the jury is still out on exactly what is going on - but it there is still an important principle that must be kept in mind. The existence of diversity within a trait is a strong indication that such diversity is necessary - no one expression of that trait is an unqualified benefit to all individuals all of the time. Otherwise, we would see much less diversity. How much diversity do we see with intelligence? A great deal. Why might lower intelligence sometimes be advantageous? We don't know - but finding out will be interesting.