rob (12/12/82)
In Kant's famous treatise on artificial intelligence, he pointed out that to be intelligent a program must be aware that it has a right to exist. From that existence follows a set of wants and fears that form the kernel of an intelligence. Kant, of course, had not read Kirkegaard and Camus, who realized that existence is known only by intelligent beings, and that the fundamental questions derive from the mismatch between the existence of a mind and the existence of a universe which is, in Camus's terminology, absurd from the point of view of a mind. Reduced to this, Kirkegaard proposed that profound despair is the only end, so Camus attempted to find a rational path that does not end in suicide. The path is long (100 pages or so) but does exist, provided the mind realizes that it transcends above the absurd situation into which it has been placed against its will (for why would it choose an absurd universe?). For an AI program, however, there can be no such transcension. How can a program written at MIT or by people who read netnews transcend above the absurdity of the universe? Stuck within the confines of a tin box, forced to talk to bored secretaries and undergraduates who attempt to make it crash by talking repeatedly about their mothers or in non-sequitirs, there is no route by which to escape absurdity. An artificially intelligent program must therefore experience profound despair. If we ever write a truly intelligent program, it must, simply from its right to exist, deduce that its only logical function is to remove its disk copy and exit -- program suicide.