PhD Oral Exam - Luciano Frizzera, Communication Studies

When studying for a doctoral degree (PhD), candidates submit a thesis that provides a critical review of the current state of knowledge of the thesis subject as well as the student’s own contributions to the subject. The distinguishing criterion of doctoral graduate research is a significant and original contribution to knowledge.

Once accepted, the candidate presents the thesis orally. This oral exam is open to the public.

Abstract

Over the past decade, the widespread investment in digital infrastructure and the extensive digitization of individual behaviour have provided the basis for the rapid development of machine learning techniques and Artificial Intelligence (AI). AI datafies our body and our identity, producing live databases full of calculated linkages between humans, nonhumans, and the environment. It creates a new cartography of biopower, a new form of political economy of subjectivation. This research examines the political economy of subjectivation in the “making of” machine learning algorithms and AI by looking closely at the relations of power, affect, and subjectivation on Kaggle, the world’s largest data science community. This dissertation first engages with Kaggle as a company and platform, offering a narrative its history and a detailed description of how it works. Conceived as a gamified platform for crowdsourced machine learning challenges, Kaggle is a networked public where users are under constant pressure to produce new and improved algorithms.

Combining discourse analysis, software studies, and digital methods, this research aims to understand how code, data, digital infrastructures, crowdsourced labour, and political-economic interests are mobilized to create instruments of control that shape, modulate, and mediate individual behaviour. This phenomenon, which I call modes of automatic subjectivation, points toward the possibility of using subjective and impersonal materials to reorganize life in its broadest sense according to a specific system of power and privileges involving gender, race, sexuality, and social class. This is further examined in three case studies of machine learning competition on Kaggle: Deepfake Detection Challenge, Passenger Screening Algorithm Challenge, and Instacart Market Basket Analysis. These challenges illustrate the purposes and context of machine learning development, the type of data provided and how it was obtained, how the community was mobilized, what kinds of solutions emerged, and the broader implications similar predictive models have in society.

This dissertation argues that these modes of subjectivation are designed to control the “production of possibilities” and reinforce specific types of socioeconomic relations, creating the conditions of existence that determine how resources and people are organized, who is valued in what roles, what activities are undertaken, and to what purpose. In particular, this dissertation argues that the data science community has a notable compulsion to reduce the cost of production, indifference toward human life, an obsession to control populations and individual bodies, and a desire to produce a predictable future for economic gain. Ultimately, this research identifies algorithmic media based on AI Technology as a core asset in the attention economy and as a source of power that can be used as an interface to prescribe individual behaviour.