Reflections
Taken from Lem’s collection A Perfect Vacuum; Perfect Reviews of Nonexistent Books, “Non Serviam” is not just immensely sophisticated and accurate in its exploitation of themes from computer science, philosophy, and the theory of evolution; it is strikingly close to being a true account of aspects of current work in artificial intelligence. Terry Winograd’s famous SHRDLU, for instance, purports to be a robot who moves coloured blocks around on a table top with a mechanical arm, but, in fact, SHRDLU’s world is one that has been entirely made up or simulated within the computer — “In effect, the device is in precisely the same situation that Descartes dreads; it’s a mere computer which dreams that it’s a robot.”[28] Lem’s description of computer-simulated worlds and the simulated agents within them (worlds made of mathematics, in effect) is as accurate as it is poetic—with one striking falsehood, a close kin to falsehoods we have encountered again and again in these tales. Lem would have it that thanks to the blinding speed of computers, the “biological time” of these simulated worlds can be much faster than our real time—and only slowed down to our pace when we want to probe and examine; “…one second of machine time corresponds to one year of human life.”
There would indeed be a dramatic difference between the time scale of a large scale, multidimensional, highly detailed computer simulation of the sort Lem describes and our everyday world’s time scale—but it would run in the other direction! Somewhat like Wheeler’s electron that composes the whole universe by weaving back and forth, a computer simulation must work by sequentially painting in details, and even at the speed of light quite simple and façadelike simulations (which is all that artificial intelligence has yet attempted to produce) take much longer to run than their real life inspirations. “Parallel processing”—running, say, a few million channels of simulation at once—is of course the engineering answer to this problem (though no one yet knows how to do this); but once we have worlds simulated by millions of channels of parallel processing, the claim that they are simulated rather than real (if artificial) will be far less clear. See “The Seventh Sally” (selection 18) and “A Conversation with Einstein’s Brain” (selection 20) for further exploration of these themes.
In any case, Lem portrays with uncanny vividness a “cybernetic universe” with conscious software inhabitants. He has various words for what we have often called “soul.” He refers to “cores,” “personal nuclei,” “personoid gemmae,” and at one point he even gives the illusion of spelling it out in more technical detail; “a coherent cloud of processes… A functional aggregate with a kind of “centre” that can be defined fairly precisely.” Lem describes human—or rather, personoid—consciousness as an unclosed and unclosable plan for a total reconciliation of the stubborn contradiction of the brain. It arises from, and “soars and flutters” over, an infinite regress of level-conflicts in the brain. It is a “patchwork,” “an escape from the snares of Gödelization,” “a mirror whose task it is to reflect other mirrors, which in turn reflect still others, and so on to infinity.” Is this poetry, philosophy, or science?
The vision of personoids patiently awaiting a proof of the existence of God by a miracle is quite touching and astonishing. This kind of vision is occasionally discussed by computer wizards in their hideaways late at night when all the world seems to shimmer in mysterious mathematical harmony. At the Stanford AI Lab late one night, Bill Gosper expounded his own vision of a “theogony” (to use Lem’s word) strikingly similar to Lem’s. Gosper is an expert on the so called “Game of Life,” on which he bases his theogony . “Life” is a kind of two-dimensional “physics,” invented by John Horton Conway, which can be easily programmed in a computer and displayed on a screen. In this physics, each intersection on a huge and theoretically infinite Go board—a grid, in other words—has a light that can be either on or off. Not only space is discrete (discontinuous) but time is also. Time goes from instant to instant in little “quantum jumps.” The way the minute hand moves on some clocks—sitting still for a minute, then jumping. Between these discreet instants, the computer calculates the new “state of the universe” based on the old one, then displays the new state.
The status at a given instant—nothing further back in time is “remembered” by the laws of Life-physics (this “locality” in time is, incidentally also true of the fundamental laws of physics in our own universe). The physics of the Game of Life is also local in space (again agreeing with our own physics); that is, passing from a specific instant to the next, only a cell’s own light and those of its nearest neighbours play any role in telling that cell what to do in the new instant. There are eight such neighbours—four adjacent, four diagonal. Each cell, in order to determine what to do in the next moment, counts how many of its eight neighbours’ lights are on at the present moment; If the answer is exactly two, then the cell’s light stays as it is. If the answer is exactly three, then the cell lights up, regardless of its previous status. Otherwise the cell goes dark. (When a light turns on, it is technically known as a “birth,” and when one goes off it is called a “death”—fitting terms for the Game of Life.) The consequences of this simple law, when it is obeyed simultaneously all over the board are quite astonishing. Although the Game of Life is now over a decade old, its depths have not yet been fully fathomed.
The locality in time implies that the only way the remote history of the universe could exert any effect on the course of events in the present would be if “memories” were somehow encoded in patterns of lights stretching out over the grid (we have earlier referred to this as a “flattening” of the past into the present). Of course the more detailed the memories, the larger the physical structures would have to be. And yet the locality in space of the laws of physics implies that large physical structures may not survive —they just disintegrate!
From early on the question of the survival and the coherence of large structures was one of the big questions if Life, and Gosper was among the discoverers of various kinds of fascinating structures that, because of their internal organization, do survive and exhibit interesting behaviours. Some structures (called “glider guns”) periodically emit smaller structures (“gliders”) that slowly sail off toward infinity. When two gliders collide, or, in general, when large blinking structures collide, sparks can fly!
28
Jerry Fodor, “Methodological Solipsism Considered as a Research Strategy in Cognitive Psychology” (see “Further Reading”).