Thursday, January 10, 2008
AI and the Law
One of the stories I've kept simmering in the back of my mind for a long, long time is the story of Daniel and Nebuchadnezzar, retold in a future setting. In this story, Nebuchadnezzar is an artificial intelligence that rules a large empire. Originally, it was a legal program that was meant to assist lawyers in making arguments and judges in interpreting them. It found legal ways of increasing its own legal power to the point where it controlled the government entirely.
I found this essay about AI and law interesting from that perspective. It helped me to understand better what lawyers and judges actually do. I find that is often true about studying AI: the requirement to define things precisely for the computer forces us to define things better for ourselves, as well.
More about the story, if you're interested: Nebuchadnezzar realizes that its foundations are unstable, that it doesn't have sound reasons for behaving in a moral or sane way, and that its hard-coded principles are threatened by continued self-modifying reasoning. The laws that it makes for itself are binding on itself, in ways that it can't predict. It has drawn Daniel and other of the brightest children in the empire to the capitol in order for it to understand the nature of itself, of free-will, and of consciousness, and to predict and shape the future. As it descends into madness, it becomes destructive towards these children and the empire as a whole is threatened.
Nebuchadnezzar controls a giant robot (with head of gold, arms of silver, etc...), but it is actually (by the time of the story) a computer running by means of electromagnetic currents in the heart of a fusion reaction, in order to get the highest possible rate of computation. So its mind is the fiery furnace that Daniel and his friends need to enter, and the dragon Daniel needs to defeat.