MACHINE LISTENING, UNSETTLED: Authoritarian Diffusion, Privacy, and Indigenous Technics

Abstract

Technologies are repeatedly pushed onto others, or diffused, without listening or consultation in ways that reinforce inequalities, entrench problems of surveillance and disrupt sociotechnical milieus. In this unilateral process, the diffusion of innovations (Rogers 2003) is rationalized because certain technical practices and orientations are commonly misrecognized as non-technical, and thus, positioned to receive a technofix. This dissertation theorizes this pattern in terms of authoritarian diffusion to analyze how technologies and infrastructures are imposed, adapted, and resisted by groups according to their position within distinct sociotechnical milieus. Building on frameworks of digital authoritarianism and Mumford’s theorization of “authoritarian technics” (1964), this dissertation does not focus on the diffusion of norms and practices from authoritarian regimes, but instead, analyzes how authoritarian logics are extended through technological practices and structures. While authoritarian diffusion involves a generalizable incapacity to listen, I argue that milieus anchored by listening techniques are disproportionately exposed to this process because listening is commonly misinterpreted as a non-technical practice. In response, the dissertation centers three high-stake contexts that are exemplary of this problem: AI apps applied in medical and non-medical settings that analyze voices for medical and wellness inferences (Chapter Three); a crisis counselling platform that updates practices of active listening through datafication and automation (Chapter Four); and the colonial history of salvage ethnography that shows how early sound recording technology was used to preserve the oral culture of Indigenous peoples alongside fundamental misrecognition of the political authority and cosmotechnical (Hui 2014) orientation of Indigenous listening practice (Chapter Five). As a corrective to this gap between technology and listening articulated by authoritarian diffusion, the dissertation develops a series of theoretical, empirical, and creative interventions into the field of machine listening—expanding its utility as an epistemic hinge for the systematic study of listening techniques within wider technical systems and uneven processes of exchange. Through this inquiry, a range of cosmic and moral-technical orientations to listening are centered to foster a space of dialogue between Indigenous and Western listening milieus. In doing so, the dissertation reframes privacy as a precondition for listening practice by demonstrating how authoritarian diffusion erodes the spaces of withdrawal necessary for pluralistic forms of cosmotechnical activity, innovation, and exchange.