The Knowing Subject in AI

• Published in: Artificial Knowing pp. 57 - 81
• Main Author: Adam, Alison
• Format: Book Chapter
• Language: English
• Published: United Kingdom Routledge 1998; Taylor & Francis Group
• Edition: 1
• Subjects: Philosophy: epistemology & theory of knowledge; Philosophy: metaphysics & ontology; Mainstream Epistemology; Traditional Epistemology; Expert Systems; AI Community; Knowing Subject; Multiple Models; AI Programming Language; Epistemic Standards; CMU; Human Problem Solving; Feminist Epistemology; Foley's Work; Symbolic Ai; Twin Earths; Epistemically Rational; Cryptarithmetic Problems; UK Press; Young Men; Consensual Knowledge; Soar System; AI System; Physical Symbol System Hypothesis; Global Knowledge Base; Gps System
• ISBN: 9780415129626; 041512963X; 0415129621; 9780415129633
• DOI: 10.4324/9780203005057-4

Chapter two examined important features of the historical development of symbolic AI, emphasizing the way in which AI reasoning techniques were developed from highly constrained and artificial problem solving situations. A number of key issues for a study of AI, partly philosophical, partly sociological, emerge from a reading of the literature on debates surrounding AI. In this chapter I want to bring together these historical considerations and key issues into a discussion of the relevance of feminist epistemology for a study of AI, and especially to see how far two AI systems-Cyc and Soar-stand as examples of the concerns I have raised. Chapter one introduced the way in which feminist epistemologists have developed a critique of rationalist epistemology, with its reliance on the 'S knows that p' formulation of who the knower should be, and what is to count as knowledge. In this and the following chapter I unpack that formulation with the aim of showing how the traditional rationalist view of knowledge permeates the design of AI systems. In particular, in this chapter I focus on the 'S' while chapter four examines p'. I argue that the incorporation of ideals from traditional rationalist epistemology has important ramifications for AI systems, not so much in the way they are used-for my example systems are somewhat removed from public use as yet-but rather in what they say, albeit in an indirect way, on the nature of knowers and knowledge.