Posts
Comments
Let me suggest that the difference between goal-less behavior and goal-driven behavior, is that goal-driven behavior seeks means to attain its end. The means will vary with circumstances, while the end remains relatively invariant. Another indication of goal-driven behavior is that means are often prepared in anticipation of need, rather than in response to present need.
I said "relatively invariant" because goals can be and often are heirarical. An example was outlined by Maslow in his "A Theory of Human Motivation" in the Psychological Review (1943). Maslow aside, in problem solving, we often resort to staged solutions in which the means to a higher order goal become a new sub-goal and so on iteratively -- until we reach low level goals within our immediate grasp.
A second point is that terms such as "purposeful" and "goal-seeking" are analogously predicated. To be analogousley predicated, a term is applied to differnt cases with a meaning that is partly the same and partly different. Thus, a goal-seeking robot is not goal-seeking because it intends any goals of its own, but because it is the vehicle by which designer seeks to effect his or her goals. In the parable, if the goal was the destruction of blue uniformed enemies, that goal was only intended by the robots creators. Since the robot is an instantiated means of attaining that gosl, we may speak, analogously, of it as having the same goal. The important point is that we mean different things in saying "the designer has a goal" and "the robot has a goal." Each works toward same end (so the meaning iis partly the same), but only the designer intends that end (so the meaning iis partly different). (BTW this kind of analogy is an "analogy of attribution.")
The fact that the robot is ineffective in attaining its end is a side issue that might be solved by employing better algoritms (edge and pattern recognition, etc.) There is no evidence that better algorithms will give the robot intentions in the sense that the designer has intentions.