Machine Habitus: Toward a Sociology of Algorithms

There is a race within the technology field to develop machines and AI technologies that mimic humans. Artificial intelligence (AI) and automated systems are transforming human lives and becoming part of our social lives, radically changing the world. Automated systems determine financial transactions, credit risks, labor, hiring, and advertisement, and they recommend what to purchase or watch next. However, while AI has benefits, there is a growing concern about algorithmic discrimination and harm. Scholars and practitioners within the interdisciplinary fields of artificial intelligence, ethics, and society have shown the harms and benefits of AI and society by focusing on biases and discrimination, fairness, accuracy, and societal impacts of the algorithmic systems. For non-technical scholars, understanding AI’s complex and technical aspects can be intimidating and challenging. In Machine Habitus: Toward a Sociology of Algorithms, Massimo Airoldi has taken up this challenge by providing the sociological tools and theories required to study the social implications of algorithms and AI technologies. After all, machines are sociological objects that affect daily lives and hold societies together. Airoldi poses timely sociological questions about AI and society and provides significant theoretical contributions to the new field of the sociology of algorithms. Throughout the book, Airoldi investigates machine learning, algorithms, and AI, which are all automated systems, introducing the reader unfamiliar with these technologies to the complex terms used to explain AI. A necessary addition to the sociology of AI, the book expertly ties together concepts from cultural sociology, computer science, AI research, and Science and Technology Studies. In five chapters, Airoldi provides detailed explanations and examples of algorithmic systems and the problems of bias and inequality. Airoldi builds his work on the classical theoretical framework of Pierre Bourdieu, specifically the concepts of habitus, agents, social fields, structure, and culture to explain contemporary social issues. The book is an inspiration for readers interested in applying Bourdieu’s sociological theory within the techno-social world. For Airoldi, machine habitus is a key mechanism where socialized algorithmic systems reproduce cultural dispositions and social structures. Throughout the book, Airoldi focuses on two key sociological questions: how algorithms are socialized—what he terms the social shaping of algorithms or culture in the code—and how the socialized machines participate in society and reproduce it—the code in the culture. First, Airoldi effectively establishes how culture shapes the codes, how machine learning tools learn from society and, specifically, culture. According to Airoldi, culture in the code occurs when machine learning systems are developed and socialized from user-generated data, design features, and decisions created by machine creators. As such, human behavior in the social world is spread and (re)shaped by algorithmic systems. Airoldi explains how through digital practices and patterns, users unknowingly contribute to the data used to ‘‘train’’ the algorithmic systems developing the machine habitus. Second, Airoldi claims the study of algorithmic systems has a sociological relevance, similar to human socialization—the internalization and learning process of the culture, language, knowledge, and social roles— machine learning algorithms also go through the socialization process. Supervised by their creators, machine learning algorithms are shaped by society, becoming social agents. While algorithms have some benefits, they can lead to bias and discriminatory behavior, mainly when the data inputs and statistical