Searching for Conciousness: Artifical Life
More and more, the lack of intentionality and ability to act has been recognized as a barrier to the development of artificial intelligence, and this has been partially responsible for the formation of a new discipline called "artificial life." Artificial life programs like Thomas S. Ray's Tierra seek to give their artificial intelligences reason to live, in essence by creating for them a virtual ecosystem. As he describes his project:
This system [Tierra] results in the production of synthetic organisms based on a computer metaphor of organic life in which CPU time is the "energy" resource and memory is the "material" resource. Memory is organized into informational patterns that exploit CPU time for self-replication. Mutation generates new forms, and evolution proceeds by natural selection as different genotypes compete for CPU time and memory space. (11)
"This evolution in a bottle may prove to be a valuable tool for the study of evolution and ecology,"(12) Ray concludes. Ray takes a purely software approach to artificial life; others use hardware (robots) or wetware (biological components.) "Although each of these areas has its distinctive emphases and research agendas," N. Katherine Hayles says in her book How We Became Posthuman: Virtual Bodies in Cybernetics, Literature, and Informantics, "they all share the sense of building life from the 'bottom up'."(13) Later, she notes that the artificial life community regards consciousness as an epiphenomenon, a latecomer to the game of life.
Whereas AI envisions cognition as the operation of logic, AL sees cognition as the operation of nervous systems; AI starts with human-level cognition, AL with insect- or animal-level cognition; in AI, cognition is constructed as if independent of perception, whereas in AL it is integrated with sensory/motor experiences.(14)
Artificial life, then, is coming from the opposite direction than the thinking which produced A.L.I.C.E, a direction which shows some superficial promise. Intentionality and perception are being given their due. However, though AL is yet a young science, there do appear to be some major difficulties. First, as Hayes points out, its usefulness as a tool for investigating natural processes is limited. Artificial life experiments hope to prove that a few simple causes, like survival of the fittest, lie at the root of even the most complex behavior, but to prove it they are building programs which already operate only on those few simple causes. They are, in essence, building their assumptions into the beast, and then pruning away those parts of their simulated or laboratory ecosystems which cause problems in the name of "debugging the system." They are looking for emergent behaviors, which they hope will in time include consciousness, but as Hayes again points out, "To create successful Artificial Life programs, it is not enough to create just any emergence. Rather, the programmer searches for a design that will lead to second-order emergence [emergences that bestow additional functionality on the system]. Once second-order emergence is achieved, the organism has in effect evolved the capacity to evolve [Italics Hayes's]."(15) Programmers of artificial life have made the same mistake as A.L.I.C.E's programmer of building their own intentionality into the system, rendering it capable only of carrying out their intentions, not its own - perhaps in some very unusual or unexpected ways, but their intentions nonetheless.
A second problem which has already been touched on is the oversimplification of the artificial ecosystem. It is theoretically possible to create a simulation which is just as complex as the simulated world, but this is very far from being practically plausible. Current simulations, whether electronic or physical, are, by definition, simplified versions of whatever ecosystem they are simulating. This does not render simulations useless; only limited. "The problem comes," says Hayes, "when this mode of operation is taken to be fully representative of a much more complex reality and when everything that is not in the simulation is declared to be trivial, unimportant, or uninteresting."(16)
A third problem - Hayes's focus - is that artificial life still privileges information processing over everything else, a view which naturally leads to the trivialization of embodiment (as information is no longer seen as physical) as well as the trivialization of human experience. "If the name of the game is processing information," says Hayes, "it is only a matter of time until intelligent machines replace us as our evolutionary heirs."(17) The focus of artificial life has been almost exclusively on the acceptance of input, with a few programmed responses to that input serving as the artificial being's intentionality. It is possible that human behavior is nothing but a system of reflexive behaviors produced in response to information input, a simple system grown confusingly complex over time, but if so, it has certainly developed far beyond anything seen in artificial life to this point. Artificial life programs can run through millions of generations of evolutionary mutations in a year(18), but none have shown anything more than very basic animal intelligence, and certainly none have shown anything resembling consciousness.
A fourth, final, and possibly fatal problem is the underlying assumption that by forcing artificial life to emulate the objectively observable behavior of biological life through its programming we can create the essentially subjective mental states, such as consciousness, that are possessed by biological life. To use the language of genetics, programmers believe that by recreating phenotype, they will force the genotype to come into existence. I suspect that this assumption is faulty. For one thing, they are trying to rebuild the phenotype and force into existence the genotype on an entirely different machine. It seems as likely that the exact same "program" of consciousness, even if such a thing existed, could be run on my neurons and a computer's chips as that one could extract energy from both dry wood and uranium atoms with a match.
previous page
next page
Contents copyright © K. Feete, 2002. All rights reserved.