(Georgia Institute of Technology, 2022-08-01)
Woo, Seokkyun Joshua
This dissertation examines “failures” across three different dimensions of the production of science (production of data, impacts, careers) to further expand our understanding of scientific work, thereby providing effective implications for science policy. The first study (Chapter 2) involves ethnographic observation of the work of bench scientists at material science labs to understand the problem-solving activities involving frequent interruptions in producing experimental data. The second study (Chapter 3) expands our understanding of citation practice in scholarly communication. In doing so, I examine citations to retracted references to test existing theories and propose an additional mechanism for how scientists embed other scientists’ works into their papers. The last study (Chapter 4) addresses the long-standing issue of gender inequality in scientific careers. In doing so, I ask how the increasingly bifurcated production role in science may shape career longevity and how this relationship may differ between women and men scientists. Together, these studies use a sociology of work perspective to better understand various components of the production of science in order to develop a deeper understanding of the science of science as well as to inform policy debates and other initiatives designed to improve the production of science.
(Georgia Institute of Technology, 2022-06-02)
Schiff, Daniel S.
As research and adoption of artificial intelligence (AI) has significantly advanced in the early 21st century, determining how to govern AI has become a global priority. Key questions include how AI should be understood as a policy domain, which policy problems are most pressing, which solutions are most viable, and who should have a say in this process. This dissertation seeks to provide key insights into the early years of AI policy, focusing on the development of the emerging AI policy agenda in the United States. To do so, it examines and reveals which issues, actors, and influence efforts are playing a prominent role in the complex, ambiguous, and contested process of agenda-setting. The research performed draws on a variety of quantitative and qualitative methodologies, including document analysis, text-as-data and time series approaches, and experimental techniques. Data examined include text from U.S. federal AI policy documents, traditional and social media discourse from federal policymakers, media, and members of the public, and engagement data collected from state legislators who participated in a field experiment.
The results reveal that social and ethical dimensions of AI receive a heightened degree of attention in AI policy discourse. However, consideration of these issues remains partially superficial and subsumed into concern about AI's potential for economic innovation and role in geopolitical competition. Further findings demonstrate that policy entrepreneurs can use persuasive narratives to influence legislators about AI policy, and that these narratives are just as effective as technical information. Finally, despite pervasive calls for public participation in AI governance, the public does not appear to play a key role in directing attention to AI's social and ethical implications nor in shaping concrete policy solutions, such that the emerging AI agenda remains primarily expert-driven. The dissertation's findings and theoretical and methodological approaches offer key contributions to policy process scholarship and related fields of research, and provide a baseline on which to understand the evolution of the AI policy agenda and AI governance going forward.