(Properties of) intelligent systems as (properties of) concrete knowledge structures
The essential point of systems based on semantic nets, frames, internal models, prototypes, and “common sense” is that, except in very special circumstances, the intelligent performance of a system depends more directly on the particular interconnectedness and organization of its knowledge, than on any reason- ing or processing power. Note that the issue is not the quantity of information, but its specific quality—its concrete structure. Everything is in the details—that these items are grouped together, that there is a cross-reference from here to there, that this topic appears in that index with those keys—in such a way that abstracting the form of the knowledge from its content would make no sense. In other words, according to this trend in AI, to study (or build) intelligent systems is, above all, to study (or build) large, concrete knowledge structures—not simple processors.
—Haugeland, J., 2000: Mind Embodied and Embedded, pp. 3-4. Chapter 9 in Haugeland, J., 2000: Having Thought: Essays in the Metaphysics of Mind. Harvard University Press.
author:haugeland-john book:haugeland-having-thought paper:haugeland-mind-embodied-and-embedded snip:intelligence-as-structured-knowledge information-architecture list:mind-body-world
