AI and the Notion of Progress (Part 3): Through the Wall

Dateline: June 28, 1998

In Part 1 of this series, I reviewed Richard Dawkins’ The Extended Phenotype, which presented evidence that genes can control not just their "own" organisms but also—from a distance—the broader local environment, including "foreign" organisms. Dawkins wanted us to view the organism in a new light—as a member of a network that has emerged through the evolutionary process to ensure the survival of certain genes. I view the extension of the genotype through the phenotype as exemplary of progress in evolution.

In Part 2, we summarized the position of one of the more strident representatives of the evolutionary non-progress persuasion, Stephen J. Gould, who would apparently deny that the emergence of anything in evolution is a sign of progress but is, rather, simply a random variation within a closed system. His closed system of evolution is bounded by a left wall, which represents the beginning of evolution, and a right wall, which limits the extent to which life can vary. In my understanding of the implications of Gould’s analysis, it seems that we—Homo sapiens sapiens—are pretty much jammed up against the right wall, and have nowhere to go.

But Gould, you will recall, left himself a loophole by acknowledging "a general and driven trend to technological progress" (his emphasis) on the grounds that (1) cultural "traditions," unlike biological species, can interbreed, and (2) that the inheritance of ideas in culture is Lamarckian (versus the Darwinian natural selection of species).

In this final part of my notes on AI and progress, we go where Gould fears to tread: through the loophole in his right wall. We will not be the first. The late eminent biologist Julian Huxley (one of the chief architects of the Modern Synthesis of Darwinian natural selection with Mendelian genetics) was there well before us, defining progress as "increased control over and independence of the environment" (quoted in Connie Barlow’s Evolution Extended: Biological Debates on the Meaning of Life. Cambridge, Mass.: MIT Press, 1994.) Culture and its ideas—in particular, its scientific and technological ideas—are indisputably the keys to our control over and independence of the environment.

Computers, robots, and AI are designed to further our control and independence, but at a distance and vicariously. Through the Mars robotic rover Sojourner, we have been able to explore the hostile Martian environment without worrying about the low temperature and absence of water and oxygen. We may one day pitch camp on Mars, but Sojourner has already been there, done that. It follows that an advanced Sojourner—one with self-consciousness and free will—would represent a species (Machina sapiens) with more control over and independence of the environment than Homo sapiens. It represents evolutionary progress, though not biological nor necessarily human evolutionary progress.

Huxley pointed to a second criterion of future progress: human values (which, by definition, cannot have been operative in evolution until humanity evolved). Values give us a sense of purpose, and having a sense of purpose leads us work toward fulfilling it. In other words, it leads to progress. To accept this, however, puts us back inside Gould’s box, on the wrong side of the wall. Our fundamental values have not changed since the beginning of recorded history, and probably since even before that.

But J.B.S. Haldane, another eminent biologist, argued that because modern wo/man "is probably an extremely primitive and imperfect type of rational being" it would be dangerous, to say the least, to look to our values for pointers to evolutionary progress. I side with Haldane on this, and have argued in a previous feature for the emergence of a higher set of values with the emergence of an evolutionary lifeform (Machina sapiens) more advanced than Homo sapiens. But to the extent that we can learn (or be taught) to communicate with Machina sapiens, then we can perhaps learn its higher values, and thus satisfy Huxley’s criterion without negating Haldane’s argument—and without jumping back in the box.

Huxley accepted that "We men are from one point of view mere trivial microbes" (feminists can have a field day with that!), but from another we are "the crown of creation" and are "engaged upon . . . the task of imposing mind and spirit upon matter and outer force."

This he does by confronting the chaos of outer happenings with his intellect, and generating ordered knowledge; with his aesthetic sense, and generating beauty; with his purpose, and generating control of nature; with his ethical sense and his sense of humour, and generating character; with his reverence, and generating religion. I’d be exceedingly hard pressed to put it better myself, but I would add: With his curiosity, and generating the next step in evolution: Machina sapiens. AI is the technologically progressive loophole allowed by Gould. Once we go through it, we—maybe not as biological organisms, but still as expressions of the information coded in our genes—will be at the left wall of a new box, with a right wall not yet in sight. I find that very exciting.
 
 

  Until next week, 

 

NEXT WEEK: Emotional Machines: We’ll get away from the philosophy and back to the science of AI with a  review of Rosalind Picard’s Affective Computing (1997, MIT Press).

Help Wanted: Got questions or comments on this article or on any other AI-related subject under the sun? Post it in the AIBB!

Previous Features