Sunday, February 7, 2010
Dialog with a Robot
I like the thought of a machine—specifically, a robot—that could explain itself down to the procedure it is executing. I imagine a machine that is able to refer to the software that defines the non-physical aspects of its self. The software driving the machine at any given moment might not be free of error. I imagine an interactive operating system to facilitate communications and that provides a basis for humans to reason about and with the software driving the machine. To illustrate this, consider such a machine named Robot engaged in a dialog with some human named Human. Let Robot and Human be at the point in a discussion when Human inquires Robot with "Why?". This is in response to Robot's previous assertion which was not understood by Human. Robot explains its meaning using a natural language that Human can readily interpret, such as English. Despite Human's familiarity with the language, Human does not understand Robot's motivation. Human again inquires, "Why?". This dialog between Robot and Human continues until Robot states "I say this due to my programming in the procedures named <Procedure A>, <Procedure B>, ..., and <Procedure N>." This seems like an appropriate behavior for machines that humans can interact with in some way similar to how one human interacts with another human because it allows such natural interaction to take place, but provides the option to the human (or perhaps another machine) to discover the precise mechanisms that entailed the behavior that summoned inquiry.
Subscribe to:
Post Comments (Atom)
0 comments:
Post a Comment