Note even though all the robot’s intentional states and acts are derived from your purposes, they are beginning to become detached somewhat from your purposes. Since you designed the robot to “think for itself” to some degree, its “thinking” may escape your anticipated boundaries. For a real-world, nonfictional example of such an artifact, consider a chess-playing computer that can beat its creator at chess. It is true that the only reason we can say that the computer is currently “investigating” queen-side rook options and “deciding” not to castle is that it is an artifact designed by a human artificer to do just that sort of thing. We are artifacts, designed over the eons as survival machines for genes that cannot act swiftly and informedly in their own interests. Our interests as we conceive of them and the “interests” of our genes may well diverge—even though were it not for our genes’ “interests,” we would not exist. Their preservation is our original raison d ‘être, even if we can learn to ignore that goal and devise our own summum bonum, thanks to the intelligence, the capacity to learn, that our genes have installed in us. So our intentionality is derived from the intentionality of our “selfish” genes.
Philosopher Daniel Dennett's Book Intuition Pumps