Editor’s note: This is a co-authored post by Bonnie Nardi and Hamid Ekbia.
For the last several years, we have tried to understand how digital technology is changing labor. Of all the alleged causes of disruptions and changes in employment and work—immigrants, free trade, and technology—the last one has received the most extensive debate lately. We review the debate briefly and then discuss our research and how it bears on the questions the debate raises.
Recently, the McKinsey Global Institute (MGI) released a report, “Harnessing automation for a future that works,” and, based on this analysis, NPR’s Marketplace launched a series on robot-proof jobs. The technology at the center of the debate is Artificial Intelligence (AI)—an area of computing research and practice with a long history of hype and false promises. This history and the confusion surrounding AI make it difficult for an outside observer to make an informed judgment about the situation. The Secretary of the Treasury Steve Mnuchin, for instance, recently dismissed the whole issue of AI replacing American workers as something that might happen in a distant future, prompting the former Secretary of the Treasury Larry Summers to criticize him in his Financial Times blog through the following analogy: “Mr. Mnuchin’s comment about the lack of impact of technology on jobs is to economics what global climate change denial is to atmospheric science or what creationism is to biology.” How can the average person make sense of the issues when prominent economists and policy makers disagree with each other to this degree?
The MGI report provides a good start for addressing this question. Extensively researched, the report seeks to rank different occupational categories on the basis of their susceptibility to automation, dividing basic capabilities into five categories: sensory perception, cognitive capabilities, natural language processing, social and emotional capabilities, and physical capabilities. Through an examination of “currently demonstrated technology,” the report makes an assessment of which capabilities are available to machines, assigning an “automation potential” to each occupational category. The report concludes that about 60% of occupations have at least 30% potentially automatable activities, with some (e.g., psychiatrists and legislators) having almost zero potential, and others (e.g., sewing machine operators) having 100% potential, with the majority (e.g., managerial) falling somewhere in between.
While thorough in its analysis, the MGI report loses sight of a whole range of computer-mediated labor currently performed by human beings in support of technological systems and economic enterprises. This labor is either uncompensated or is minimally compensated. We refer to this type of labor as “heteromation,” meaning that it combines capacities of humans and machines—with an emphasis on the machines (and their owners). Heteromation includes many kinds of self-service (e.g., check your own groceries), volunteer work such as citizen science, creative engagement (e.g., in computer gaming), microwork (Amazon Mechanical Turk), writing online customer reviews, and a wide range of other activities that provide economic value to companies and organizations, but little or no monetary compensation to laborers —that is, most of us. It was in recognition of these kinds of activities that we came to coin the term heteromation—a new division of labor driven by recent changes in capitalist economies enabled by computerization.
As it happens, in our book, Heteromation and Other Stories of Computing and Capitalism (MIT Press, 2017), we also came up with five categories of labor—namely, communicative, cognitive, creative, emotional, and organizing—which seem to closely overlap with MGI’s categories. These categories are based on our own ethnographic work, as well as close readings of the literature. The apparently similar categorizations between MGI and us suggest that we should both perhaps arrive at similar conclusions. In a way, this happens when the MGI report ends with the following note:
But our analysis shows that humans will still be needed in the workforce: the total productivity gains we estimate will come about only if people work alongside machines.
Our twist on this message is that people are already “working alongside machines”—but they are not getting paid for it, or not very much. The debate is a false battle between AI and people—when in fact the critical issue is that people are working, but the terms of labor are radically changing. Seduced into (gaming, social media…) or forced to (self-service, gatekeeper apps like Academia.edu…) participation in heteromated labor, we supply a lot of it; capitalism has, remarkably, found its way to free or very cheap labor with absolutely no obligations to laborers as human beings with basic needs.
In the book, we examine many kinds of heteromated laborers, from YouTube video creators to social-robot caretakers to Mechanical Turkers. Even with abundant examples, we know we only scratched the surface, and we continually notice fresh examples. Heteromation represents an evolution of labor relations as capitalism continues to wrangle with the age-old problem of the cost of labor. Much heteromated labor involves the economic contribution of real effort, labor, and time by people who are actually doing things that were previously done by paid workers, or new work that has arisen as uncompensated, economically valuable labor, such as writing Yelp reviews, contributing political essays to sites like Daily Kos, or answering questions on forums—all of which add to the bottom line of big corporations (Amazon. com, Blizzard Entertainment, Google…), as well as being the sole basis of the business model for some companies (Yelp, Tripadvisor, Twitter…).
So, while “humans will still be needed in the workforce,” are we going to compensate them? Mechanical Turkers earn about $2.00 an hour, yet many of them need the money for basic living expenses, not as “extra” income as some economists suggest (as we discuss in the book). Will free labor or painfully cheap labor come to seem capitalism’s due? How will we handle social security, workers’ compensation, and the safety net that has been built into secure jobs, a type of employment that is fast disappearing? If people’s labor economically supports companies, can we share in governance? Can we even just be allowed some simple preferences in rigidly prescriptive software? Yelp, for example, forces users to view Yelp’s preferred reviews first, unless a setting is changed every single time. We talk a lot about personalization and customization in computing, but in practice, technologies such as machine learning (which could notice what we are doing and set the preference), and even 1990s-style forms-based user interfaces are withheld.
We don’t have answers to these questions ranging from life and death matters of making enough money to survive, to the petty corporate diktats of user interfaces, but our observations on the increasing pervasiveness of heteromation suggest that we should be thinking about them. In this light, we suggest that the terms of the debate between economists such as Summers and Mnuchin should change from whether or not AI is around the corner to whether or not human beings, their contributions, and their needs, should be at the center of economic policy.
Ekbia, Hamid, and Bonnie Nardi. 2017. Heteromation and Other Stories of Computing and Capitalism. Cambridge, MA: MIT Press.
About the Authors
Hamid Ekbia is Professor of Informatics, Cognitive Science, and International Studies at Indiana University, Bloomington, where he also directs the Center for Research on Mediated Interaction. He is interested in the political economy of computing and in how technologies mediate cultural, socio-economic, and geo-political relations of modern societies. His most recent book with Bonnie Nardi, Heteromation and Other Stories of Computing and Capitalism (MIT Press, 2017) examines computer-mediated modes of value extraction in capitalist economies. His earlier book Artificial Dreams: The Quest for Non-Biological Intelligence (Cambridge University Press, 2008) was a critical-technical analysis of Artificial Intelligence. He is the co-editor of a volume titled Big Data Is Not a Monolith (MIT Press, 2016).
Bonnie Nardi is a Professor in the Department of Informatics in the School of Information and Computer Sciences at the University of California, Irvine. She is interested in political economy, computer gaming, and the long-term sustainability of the planet (which are all related). She is a founding member of the Center for Research on Sustainability, Collapse-preparedness and Information Technology at UC Irvine, and the Computing within LIMITS Workshop, an international effort. Bonnie is the author of My Life as a Night Elf Priest: An Anthropological Account of World of Warcraft (Michigan, 2010) and a co-author of Ethnography and Virtual Worlds: A Handbook of Method (Princeton University Press (2012).