Corporate Open Science and Relational Infrastructure in Artificial Intelligence Research
Publisher
The University of Arizona.Rights
Copyright © is held by the author. Digital access to this material is made possible by the University Libraries, University of Arizona. Further transmission, reproduction, presentation (such as public display or performance) of protected items is prohibited except with permission of the author.Embargo
Release after 05/23/2029Abstract
What is most puzzling about the rapid commercialization of artificial intelligence today is that it appears to be coupled with too much openness in science. For years, the post-WWII American success in science and technology has been told as a story of a usefully porous yet carefully guarded division of labor between science and markets. Despite their historically consanguineous and now symbiotic relationship, the two totemic institutions of modern society remain definitively distinguishable by their distinct organizational forms and incongruent interests in open versus proprietary knowledge. Universities are the primary producers and brokers of open knowledge, the kind of knowledge that is supposed to be fundamental, driven by pure scientific curiosity, and made available to the whole society for the greater good. Companies, as patrons of research contracts and beneficiaries of technology transfer from universities, are primarily interested in turning knowledge proprietary and profitable; they should be agnostic, if not antithetical, to a movement to make knowledge available to all. Over the past decade of explosive growth in artificial intelligence research, companies have nonetheless defied this very boundary between open and proprietary knowledge to become themselves leading contributors to open science infrastructures such as arXiv and GitHub, which collect, store, and disseminate unprotected, open-access papers, code, software, and datasets, a phenomenon which I term corporate open science. This dissertation examines how corporate open science has become a relational infrastructure of corporate power and allowed tech companies to control the development of artificial intelligence technology by transforming the science behind it. Existing explanations, under the assumption about the post-WWII American model of institutional boundaries between science and markets, have focused on companies’ interests in proprietary forms of knowledge. This study, however, by tracing new forms of relationships that open science has allowed tech companies to forge across boundaries, along careers, and with their new labor force between 2013 and 2021, offers an alternative narrative of how and why tech companies have become leading contributors to open-access repositories of papers, code, software, and datasets such as arXiv and GitHub – the core infrastructure of open science.Type
Electronic Dissertationtext
Degree Name
Ph.D.Degree Level
doctoralDegree Program
Graduate CollegeSociology