Reviewed by: Artificial Knowing: Gender and the Thinking Machine * Bayla Singer (bio) Artificial Knowing: Gender and the Thinking Machine. By Alison Adam. London: Routledge, 1998. Pp. v+210: notes/references, bibliography, index. $75 (cloth); $22.99 (paper). Alison Adam has written a feminist polemic of limited use to historians of technology. In correctly identifying the extreme reductionism practiced by those developing “artificial intelligence” (AI) or “expert systems,” she falls into the corresponding error of overgeneralizing all the omitted aspects as “feminist.” Many readers will be uncomfortable with the style of her presentation. To quote her introduction: “I am conscious that there is, of necessity, an element of zig-zag. . . . [T]here is a fair amount of introductory material and that it is chapter three before the ‘meat course’ arrives. . . . [My book is] a Chinese banquet, made up of lots of little courses of different flavours. . . .” (p. 3). The first chapter names and describes varieties of feminist approach; the second contains an overview of AI’s history and some objections to its epistemological foundations. In the third, “meat,” chapter, Adam begins to examine two AI systems, Cyc and Soar. Her analysis continues through chapters 4 and 5, with chapter 6 presenting some suggestions for “Feminist AI Projects and Cyberfutures.” Throughout, the primary focus is feminist theory. “[I]t is the job of feminist epistemology to offer a broadside attack on traditional forms of epistemology, and to expose the ways in which women are denied the status of knowers, and what they know is denied the status of knowledge” (p. 28). “In a world where ‘expert’ almost always means white, middle-class, male experts, it is difficult to see how expert systems could contribute to the pluralistic discourse argued for by much of feminist theory” (p. 42). More than that, the experts emulated by the Cyc and Soar projects are mathematically inclined, academically gifted males. Adam convincingly demonstrates that the tasks set for Cyc and Soar are limited and highly artificial: solving mathematical puzzles rather than, say, [End Page 174] reading a newspaper with comprehension (p. 126). She further explores the “disembodied” character of “rationality” as defined and used by epistemologists and AI researchers. “[N]either Cyc nor Soar have satisfactory ways of dealing with the propositional/skills distinction. . . . This results in a very narrow conception of what it means to act intelligently” (pp. 127, 128). Although Adam gives some lip service to the fact that tacit knowledge (“know how” rather than “know what”) is involved in many aspects of technological and other forms of expertise not usually gendered feminine (e.g., pp. 12, 111), she nevertheless treats all those forms of knowledge excluded from “traditional . . . epistemology” as the proper subject of feminism rather than considering them as part of a broader critique of reductionism. In her concluding chapter, Adam begins by urging that “feminism is a political project and the best research is where action proceeds from description” (p. 156). Adam sees her work as “showing the ways in which AI can be informed by feminist theory and can be used for feminist projects.” Quoting Sue C. Jansen, Adam expects “feminist semiological guerrilla warfare . . . to transform the metaphors and models of science.” Her examples are hardly as robust as that: the first, “AI and Feminist Legal Theory,” suggests only that the expert system be “nonthreatening” to women “who have so little sense of themselves as persons with rights” that they have “difficulty in recognizing when their rights have been violated” (p. 160). In the second, “Feminist Computational Linguistics,” Adam refers to studies that have found gender differences in conversational behavior and ends by acknowledging that “the model described here is a white, middle-class, Anglo-American English one, which probably does not even fit, for example, New York Jewish speech” (p. 163). It is nowhere clear just what sort of feminist “action” Adam is proposing in this case, nor how AI expert systems might be useful to it. Adam neither promises nor delivers any historical perspective on the likely impact of their reductionist foundations on the eventual success of AI or expert systems, nor on the social influences likely to shape the product before and after it reaches the market. Her...
Search from 250 M+ research papersSearch from 250 M+ research papers
Jul 1, 2012
Dong Xiao Liu 
Open Access