-
Introduction
- The MIT Press
- Chapter
- Additional Information
Introduction Much of the argument of this book has dealt in one way or another with the relationship between technology and society. Rather than studying the social impact of technology, the authors have been more concerned with showing how technology itself can be understood as a social product, or at least as possessing a social dimension. This entails a radical shift in how we conceive of technology and the innovation process more generally. The social aspects of technology do not start when a technological process or product is taken up by the wider society; rather they are always present. This new image of technology can itself be expected to have an impact, particularly in the domain of technology policy (Wynne 1983). Understanding of the social (and for that matter the political and the economic) dimensions of technology may be crucial in understanding the success and failure of technologies in different contexts. In restricting the social, political, and economic dimensions of technology to the area of application , policymakers may have been misled. Clearly, the consequences of different models of technology for policy are an area that deserves further attention. Apart from its possible implications for technology policy, the new understanding of technology that is evolving can potentially feed back into the development of technology itself. This is one of the issues that is considered by the last two papers in this collection. They both take as their topic artificial intelligence (AI), a field that promises “to found one of the key technologies of the later part of the twentieth century” (Collins, this volume). AI researchers, in attempting to provide an account of the working of knowledge-based systems, can be seen to be engaged in a similar task to that undertaken by historians and sociologists of technology. And that is the second issue at stake in this part: how the results of analyzing AI with the newly developed tools of the history and sociology of technology feed back into that research program and into the philosophical questions 300 Technology and Beyond behind the new approaches to technology study presented in this book. The concern of AI is, of course, not merely to understand the basis of technical knowledge but to put that understanding to use by building “intelligent” machines that will be capable of carrying out human activities . Both Steve Woolgar and H. M. Collins take up the challenge of AI in their contributions, but in interestingly different ways. Steve Woolgar extends the argument of an earlier paper on the divide between humans and machines (Woolgar 1985) and shows how this boundary is socially mediated. The debate over what counts as an intelligent machine and over the likely limitations of the AI research program can be seen to be a continuation of the long-running controversy over the uniqueness of mankind. In this controversy properties that define human capabilities as unique are, according to Woolgar, shifted such that, whatever tasks machines carry out, those tasks become defined as not part of mankind’s uniqueness. The interpretative flexibility of the technology is demonstrated by this debate. Woolgar takes up Callon’s metaphor that technology, because it also involves sociology, can be used as a tool for sociological analysis. He proceeds to show how the methods of sociologists in their critique of AI and cognitivism construct a particular view of the human/machine relationship , which, in the end, imposes severe limitations on their critique. Woolgar takes the basis of the AI research program to be “cognitivism,” namely, AI’s attempts to explain behavior by reference to cognitive or mental states that can then be codified within some algorithm. The core of much of his argument is that any attempt to replace cognitivism by a systematic sociological understanding of human behavior makes little difference to the AI task because any such sociological understanding involves the delineation of human actions, thus rendering such actions capable of codification and hence of incorporation within the AI program. Collins offers a model of knowledge transfer based on his sociological studies of how a particular technical task (building a TEA laser) is accomplished . Collins argues that the “enculturational model” of knowledge, in which knowledge is equated with culture and hence has an irretrievably social element, provides a better understanding of laser building than the “algorithmic model” in which all the technical actions to be followed can be completely specified by a set of algorithms capable of transfer to a digital computer. The tacit knowledge component of technical tasks in...