For the "thinking a panda is a vulture" problem, don't humans fail in similar ways? The analogous examples for us are camouflage, optical illusions, logical fallacies, etc.
It doesn't really have to be perfect as long as it doesn't fail in common scenarios.
It doesn't really have to be perfect as long as it doesn't fail in common scenarios.