(no subject)
Rodney Brooks, "The Relationship Between Matter and Life."
A short article which I've seen referenced in a number of articles relating to embodied AI and its variants; in it, the author discusses the progress, and more importantly the lack-of-progress, in both artificial intelligence and artificial life, and speculates on what this means and what new discoveries may be required to overcome them. He does first emphasize that neither field has been barren or unproductive; he lists a number of examples, such as speech recognition programs and machine design, where AI and AL have produced powerful and widely used applications. But, clearly, pattern recognition algorithms and genetic algorithms and etc, while useful, do not approach the depth and power of "real" intelligent and living things. Something is "missing" in all current models of AI and AL (an evaluation which, though made here in 1991, is still true today.)
Brooks lists a few broad categories of what could be the limiting factor; incorrect parameters or incomplete formulations of experiments, insufficient computational power or complexity, or something "mysterious," yet to be discovered, which adds a previously missed dynamic of such systems. The first category he largely dismisses; over the long history of both disciplines, the "correct" combination of existing features would have likely been discovered by design or chance. The second category he deems plausible, citing the seemingly qualitative difference in behavior of "brute-force" chess AI programs from the early, low-powered versions to Deep Blue, and speculates that it could occur in more profound ways for AI and AL systems simply by "scaling up" the kind of models currently created until they become sufficiently powerful or unpredictable to rival real systems. The third category is the most speculative; fundamental discoveries in computer science, mathematics, or even physics may be required to completely model AI or AL systems, but currently no one can say when or if such a discovery would emerge.