Applied Epistemology

Applied Epistemology2018-07-14T15:46:28+00:00

Epistemological Engineering (EpE):  The engineering discipline through which human knowledge is modeled as a computable data or information structure within a computer as well as the development of software routines that process that structure to perform task that require knowledge and comprehension.

EpE is, like mathematics, partly discovered and partly invented. It is based on the realization that the conceptual building blocks of knowledge, that is, the “container concepts” of knowledge about real world external things, exist in a number of discrete categories and that instances within these categories may only be combined with other instances to create more complex concepts according to fixed rules.

This is highly analogous to the chemical elements which, though small in number, combine to create innumerable discrete substances with diverse physical properties.  The ability of atoms to combine and to produce identifiable categories of substances with similar properties can be predicted through a small number of core principles within atomic physics.

EpE consists of the identification of the basic categories and core concepts that underlie human knowledge of the world along with the principles that control how they may be combined to create any new concept in response to inputs from outside the system.  It is a software implementation of a relatively small number of core principles that are to Artificial General Intelligence (AGI) what the Periodic Table of the Elements is to Chemistry.

EpE does not focus on intelligence in the sense of processing raw data but rather on the product of human intelligence: knowledge of the world created by humans over the course of human civilization.  It proceeds by painstakingly implementing a small subset of this world model, roughly equivalent to the knowledge that a child possesses who has just learned to talk.  The software can then build and extend this knowledge model using well-understood and relatively simple pattern-matching and inference algorithms, layered on the sophisticated core model structure, in response to inputs from outside the system.

This hand-built core model is the key and it represents the fundamental departure taken by EpE compared to what others are doing in the field of AGI.

EpE considers knowledge a model. Models cannot be considered true or false, just better or worse at achieving the function for which these were designed. This represents a fundamental departure from traditional philosophical Epistemology which has been concerned more about truth, belief and justification than the underlying structure of knowledge.