The Technology Trap: Capital, Labor, and Power in the Age of Automation

Abstract

In 2013, Carl Benedikt Frey and Michael Osborne published a paper claiming that 47 percent of jobs in the United States were in danger of being replaced by artificial intelligence (AI). The paper made quite a splash at the time. Apparently even Barack Obama's Council of Economic Advisors took it quite seriously (322).A decade later, Frey's fears haven't quite come to pass. (A Bureau of Labor Statistics report from July 2022 in fact finds little evidence that AI has made any appreciable difference in the rate or character of job loss in the United States.) Still, out of that 2013 paper came Frey's most recent book, a survey of the history of mechanization in Europe (mainly England) and the United States. In it, he claims that we currently stand on the threshold of a “technology trap.” Until the ascent of the bourgeoisie in the eighteenth century, Frey argues, European authorities generally outlawed the introduction of machines that threatened the livelihood of craftspeople. The result was to delay industrialization and all its benefits. This was the first technology trap: when those with political power blocked mechanization, or what Frey calls, simply, “progress.”Of course, industrialization dispossessed many working people, as Frey acknowledges. It is one of his axioms that workers automatically attack machines that threaten their income—evidently regardless of the influences of culture, politics, law, and so on. As a result, machine wrecking was a “rational response” by Luddite English workers at the turn of the nineteenth century (125). With AI, Frey continues, workers once again find their incomes threatened, but with one important difference: “Unlike the situation in the days of the Industrial Revolution,” he writes, “workers in the developed world today have more political power than the Luddites did” (xiii). That is to say, while Luddites could only attack machines, working people today may attempt to regulate the rate or character of technological change through democratic governance. This, Frey tells us, would be bad, for he is adamantly opposed to “slowing down the pace of progress or restricting automation” (xiii). To oppose the degradation of working conditions wrought by “automation,” in other words, would be to slow down or restrict progress. It therefore falls to society to ameliorate the lives of those who will lose from automation, Frey holds, so that they don't short-circuit “progress” by way of that troublesome institution, democracy.I must be frank. I did not think it was still possible to write a book like The Technology Trap. Perhaps the single most important contribution to come out of the history of technology in the past half century has been its debunking of the fallacy of technological determinism, the idea that technology is a pure force of history that develops itself autonomously. Frey seems unaware of this major (and not particularly recent) insight as he narrates the history of mechanization from European antiquity to modern-day globalization. Technology is, in his words, “a mysterious force” in which “change begat change as a self-reinforcing cascade of progress created the modern world” (144, 7). Frey mentions that “social inventions”—what the rest of us might call politics—played some role in all this, but they are “beyond the scope of this book” (145, 161). This approach leads to a predictable string of outrageous historical claims: that technology itself led to the rise (and demise) of the middle class (145); that women have been liberated from domestic labor, mainly thanks to the mechanization of the home (3, 160–61); that technology created modern democracy (267). To provide just one telling example: according to Frey, the Triangle Shirtwaist Factory fire happened not because management locked workers inside a crowded sweatshop (which Frey does not mention, just as he does not mention workers’ demands for improved safety) but because of a lack of technological development. Electric lighting made the workplace safer, you see, and the Triangle Shirtwaist Factory was, sadly, not electrified. (Apparently Frey is also unaware that the fire was most likely caused not by gas lighting but by an errant cigarette.) To hear Frey tell it, electrification—not unions, or bosses, or the subcontracting system—was the only significant historical actor in this episode (193–94).None of this is new. Frey's arguments are typical of the automation discourse, a story of technological determinism that depicts employers’ control and their degradation of working conditions as synonymous with progress. The discourse hides the politics and the contingencies of history behind a seemingly neutral, technical edifice, where changes to the means of production happen on their own. When capital wins and workers lose, it's the result of a “mysterious” force. It's therefore no surprise that none of Frey's policy recommendations involve regulating (or taxing) the means of production through democratic institutions. His suggestions of wage insurance, subsidized housing, and tax credits presume that the problem ultimately lies with working people themselves rather than the structure of the economy. Frey recommends improved education as a means to help working people join the “cognitive elite” who will be needed in the ostensibly automated future. I actually had to rub my eyes when Frey called on none other than Charles Murray—of The Bell Curve—to substantiate his discussion of cognitive elitism, as when he claimed that poor people are less likely to hold undergraduate or graduate degrees because “children from low-income families often lack the intellectual stimulation from in-home reading and daily conversation that are so common in families where one or both parents have completed college” (252, 350–51).This discussion of course becomes moot when one considers that the automation “revolution” as Frey describes it may not be happening at all. Consider Aaron Benanav's research on early twenty-first-century labor productivity; or Kate Crawford's recent critique of AI as an extractive technology; or Gray and Suri's scholarship on computer “ghost work”; or Jason Smith's account of service work. Even to stay within the literature of economics, as Frey does, the Bureau of Labor Statistics recently reported that labor productivity (which in the midst of an “automation” revolution should be skyrocketing) has actually remained remarkably low for the better part of the last twenty years. Frey anticipates this criticism, explaining that low productivity “can happen when technologies are at an experimental stage” (327). In other words, according to Frey, the absence of evidence of automation is evidence of automation.Although The Technology Trap may offer little as a work of history or economics, it is not worthless. Usefully, it articulates what is usually only subtext in the automation discourse. From it, we learn that the real danger for technocrats isn't the technology trap at all but a threat of an entirely different kind: democracy.