No, the idea of neural nets is that writing smart functions doesn't scale, not that it's not possible. Take eg. Chess: the possibility space is simply too large to write a smart function for every possibility.