I was invited to give a lightning talk on my research at a White House Office of Technology Policy summit on AI: Social and Economic Impacts in the Near Term. I gave this five minute talk distilling key parts of my research that challenge some dominant assumptions by economists and policy makers — particularly The Second Machine Age by Brynjolfsson and McAfee. People have told me that the talk was helpful to them and pushed points they had missed in my work before, so I’m reposting it here despite hating the video thumbnail facial expression!
Antonio Cassilli drew out these points in a recent post:
1) “Automation doesn’t replace labor, it displaces it”: computers learns to recognize texts, images, sounds via human computation workers who fuel AI by performing micro-paid & unpaid #digitallabor on platforms like Amazon MTurk (and many more). These workers “bridge the gap between AI and changing human culture”.
2) Micro-workers face a race to the bottom in an ever-expanding, largely unregulated labor market. (Personal addendum: We cannot think AI regulatory policies without also thinking about regulating these labor markets. So next time someone asks you “what should the government do to regulate artificial intelligence to keep it ethical?”, you might wanna answer: “Let’s talk about working conditions, modes of remuneration, health care of people that do AI”.)
From the abstract: “Hackathons sometimes produce technologies, and they always, however, produce subjects. This article argues that the hackathon rehearses an entrepreneurial citizenship celebrated in transnational cultures that orient toward Silicon Valley for models of social change. Such optimistic, high-velocity practice aligns, in India, with middle-class politics that favor quick and forceful action with socially similar collaborators over the contestations of mass democracy or the slow construction of coalition across difference. “
I reviewed The Second Machine Age by Brynjolfsson and McAfee and Mindless by Simon Head for Public Books. The essay, entitled Justice for “Data Janitors” looks at the value of labor in the production of technological “magic.”
I’ve had a new piece come out in South Atlantic Quarterly as part of a great special issue on extraction, logistics, and finance. The piece is entitled Difference and Dependence among Digital Workers: The Case of Amazon Mechanical Turk. In it, I explain the implications of organizing crowdworkers as programmable “human computation,” both for the subjectivities of highly valued tech workers, the valuations of their companies, and the limits of solidarity among the multitudes.
In an essay in ACM’s interactions magazine, Six Silberman and I write about the transformation of Turkopticon from a critical design project into an infrastructure some depend on, and what that tells us about the limits of focusing on design and innovation. I was especially excited to see that this one, unlike the standard scholarly paper, has an illustrator’s interpretations of the argument!
I’m helping organize this conference. Workers, scholars, activists, organizers, we hope to bring you together around these questions. There is some funding available to support travel and attendance of digital workers! See the call
Crowdsourcing systems do more than get information work done. This paper argues that microwork systems produce the difference between “innovative” laborers and “menial” laborers, ameliorating resulting tensions in new media production cultures in turn. This paper focuses on Amazon Mechanical Turk (AMT) as an emblematic case of microwork crowdsourcing. Ethical research on crowdsourcing has focused on questions of worker fairness and microlabor alienation. This paper focuses on the cultural work of AMT’s mediations: divisions of labor and software interfaces. This paper draws from infrastructure studies and feminist science and technology studies to examine Amazon Mechanical Turk labor practice, its methods of worker control, and the kinds of users it produces.
Turkopticon was the focus of a story “Software Aims to Ensure Fairness in Crowdsourcing” in the August 2013 issue of the Communications of the ACM, the widely-read magazine of the Association for Computing Machinery. For me, coverage in the premier computing association magazine gives me a sense of triumph, as if we scaled the walls of the Computer Science (CS) profession somehow. Questions of politics and ethics are often marginal within CS departments; our hope was that Turkopticon might draw attention to the labor and power issues in computation not only among communication scholars, but also among the computer scientists building these labor systems.