DCGAN is initialized with random weights, so a random code plugged to the network would generate a totally random impression. Nevertheless, while you might imagine, the network has many parameters that we can easily tweak, and the purpose is to find a environment of those parameters that makes samples generated from random codes seem like the coach