Doctoral Dissertation Research: Encoding the Human in AI Development, Ethics, and Safety Communities

Yale University, New Haven CT

Investigators

Abstract

How do AI developers reckon with the social impacts of their technologies? This dissertation research explores the front lines of artificial intelligence, where recent advancements have provoked both excitement and concern among industry members and the broader public. By studying how developers respond to these advancements during their day-to-day work, the project documents the emergence of beliefs, norms, and values in the industry. In turn, it shows how a technology industry’s politics and social structure influence its wider benefits and harms. The project’s findings can support the work of researchers in AI safety and ethics seeking to enhance AI’s impacts. The project can also inform the activities of policy makers and activists looking to manage the effects of this emerging technology as well. Results will be disseminated through academic and public media articles, presentations to AI development, safety, and ethics researchers, and a doctoral dissertation. The project uses ethnographic methods to study the work and lives of researchers in AI development, safety, and ethics. It is focused specifically on machine learning models, and to what extent the algorithms that they depend upon are taken up ontologically, epistemologically, and ethically by those who create the AI. The project features one year of participant observation at AI companies, academic institutions, industry events, and spaces outside of work. It also employs interviews and group salon discussions with industry members and their interlocutors in business, government, and the media. The questions structuring the ethnography include: (1) how is AI development a social, cultural, and locally specific practice; (2) how do discourses about the human among AI developers configure the political stakes of AI development and deployment; and (3) what role do ethical values play in the development, critique, and artifacts of AI systems? The project extends scientific understandings of the relationship between technology, humanity, and ethics. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.