UC Berkeley

By the early 1950s, after more than a century of industrial growth, American popular culture was awash in visions of technologically advanced futures. Amidst this, a small group of computer researchers developed something they called “artificial intelligence.” In my dissertation, I examine the work of three first-generation artificial intelligence researchers, Herbert Simon, John McCarthy, and Marvin Minsky, focusing on the ways they imagined industrial progress. In so doing, I show how the architects of artificial intelligence were collectively committed—ideologically and practically—to maximalist claims about the social significance of technological change. They were convinced that the public misunderstood their work. Their response was to speculate, repeatedly and publicly, about the likely effect of artificial intelligence on history and society. In their writings, they repeatedly ventured beyond the confines of academic computing, into economics, environmental politics, the future of medicine, and much else. My dissertation examines their science and speculation alike, examining the consequences of such a joining. The four chapters of this dissertation, which focus, respectively, on the emergence of artificial intelligence as a scientific field in the 1950s, the industrial automation debates of the late 1950s and 1960s, the resource crises of the 1970s, and the “computer immortality” theories of the 1970s to 1990s, provide windows into the political valences of an advanced computer discipline. Taken together, these chapters demonstrate how artificial intelligence, as a science, could hardly be separated from broader expectations of rapid technological development. The practical work of artificial intelligence was a program of radical horizons; if successful, it would remake the shape of our lives, from how we labored, to the spaces where we lived. In this ambition it was also, I argue, the product of a historical paradox. For although artificial intelligence began its life as a technology of high-modernist planning and centralized control—as a symbol and as a tool of rationalistic decision-making—it nevertheless survived the decline of the state-centered projects for which it had been designed. In fact, it participated in their delegitimization.

Much of the historical literature concerned with industrial conceptions of space, time, and history has focused on the technologies of the Second Industrial Revolution, especially the emerging factory system and new forms of transportation and communication—in essence, a literature focused an era when rapid industrial change was still unprecedented. By contrast, the argument here emphasizes a period thereafter, in the mid-twentieth century, when technological progress had become normal, expected, and actively anticipated. Artificial intelligence, a research project with ambitious promises and a long drawn-out development, is a perfect case study for these forward-looking, anticipatory processes, for a few reasons. First, because it generated so much attention in so many quarters—academic, industrial, military, and public—it is a useful central location from which to survey the larger social forces which anticipate technological change. Second, because there was no stable agreement determined which computer programs counted as ‘real’ artificial intelligence, there was no single moment or program which amounted to its ‘invention.’ There is nothing like the flight at Kitty Hawk in the history of artificial intelligence. Computer programs which were heralded as the ‘first step’ toward artificial intelligence in one decade were demoted to ‘normal’ computer programs in the next, with ‘real’ artificial intelligence always hovering on the horizon, like the end of a rainbow. In this space of uncertainty, historical narratives of progress mattered as much as any technical detail, coloring the interpretation of artificial intelligence even for technicians.

What can be gained from viewing artificial intelligence as a site productive of historical discourse? In part, it explains the paradox above—the way artificial intelligence never seemed to designate any fixed collection of programs, but was always rather vaguely directed toward the ‘real’ and ‘true’ artificial intelligence of the future. By understanding the narrative and imaginative role artificial intelligence played in an imaginary of modern time, as anticipation, we can better understand why it was continually ‘being invented.’ Artificial intelligence was not the only technology to perform this role—automation also comes to mind—but it serves as an exemplary case study for it. In the second place, historiographically, the study’s focus on discourses of technological anticipation allows it to unite the history of computer technology with three scholarly traditions: an older, more theoretically-grounded tradition of scholarship focused on the nature of historical time and historical consciousness; a literature on the time sensibility of the second industrial revolution usually concluded by the Second World War; and a theoretical literature (largely Marxist) which is dismayed by a post-industrial world bleached of historical consciousness. This first, an almost philosophical tradition concerned with distinctly “modern” perceptions of historical time, usually hints at the important place of technological change in modern understandings of history, but rarely pursues the issue. The latter two, set in periods before the computer existed, and after history has supposedly withered, pass by the historicity of the computer altogether. By examining sites where technological progress was defined and articulated, as is done here with artificial intelligence, these discussions are given empirical foundations in the late twentieth century. In this light, artificial intelligence, which is more typically interpreted as an event in the history of philosophy or technology, is revealed to have been an important site for historical thought, albeit of an unorthodox kind. If artificial intelligence complicated longstanding philosophical issues, it was not only because it blurred the distinction between “the human and the machine,” but also because it raised questions about historical progress in an industrial society.