Badass Feminist Politics: Exploring Radical Edges of Feminist Theory, Communication, and Activism

outputs. Airoldi transforms computer scientists and developers’ ‘‘garbage in, garbage out’’ into ‘‘society in, society out’’ (p. 43), inventing a much-needed sociological explanation of algorithmic bias and discriminatory behaviors stemming from data and design of machine learning tools. Airoldi thoughtfully argues that the code in the culture occurs when socialized machines act as social agents, participating in and shaping societal and cultural practices. Socialized machines are ‘‘also more than tools; they are agents, embedded in feedback loops where machine learning and social learning compenetrate each other’’ (p. 71). In each chapter, Airoldi effectively uses examples of automated systems, such as Google’s virtual assistant that makes restaurant or hair appointments on behalf of its users. Airoldi highlights the fact that, while this feature can, for example, save time, from a sociological perspective the virtual assistant influences human social interactions. Classification systems filter and rank the social world, while recommendation systems guide users on what to buy and what movies or television shows to watch, thus becoming more influential than ‘‘human cultural intermediaries such as critics, producers, and journalists’’ (p. 83). Readers will appreciate the comprehensive range of machine learning algorithms showing how socialized machines act as social agents with machine agency and authority. Airoldi highlights how machines are involved in cultural reproduction by reshaping social interactions, relations, and the social order, raising many questions for the readers about human and machine agency. A key strength of the book is the theory of the machine habitus that Airoldi ambitiously develops. The theory’s premise is that machine habitus is the outcome of primary and secondary socializations. Airoldi creates four theoretical points that constitute the theory: these include structures—social structure and digital infrastructure; entanglements— human-machine interactions within the techno-social fields; trajectories—temporality and multiplicity, the effects of feedback loops on cultural disposition trajectories of humans and socialized machines over time and across fields or platforms; and social, symbolic, and automated boundaries. Airoldi notes that the limitations of his theory and the complex matter of algorithms ‘‘might risk producing oversimplifications’’ (p. 112) and attempts to simplify the theory using practical examples and a fictional character, Andrea, situated in real-life contexts. Perhaps a chapter using the theory with data and observations would have been a more robust addition to the book and the theory of machine habitus. Nevertheless, researchers may apply or test the theory to sociologically understand machine learning and algorithmic systems. Overall, Machine Habitus is an engaging theoretical book that provides significant insights into the socio-technical aspects of algorithms. The interdisciplinary perspectives from the book will appeal to many readers and be useful in classroom discussions. Scholars of the sociology of algorithms, technology, and culture as well as readers interested in studying the societal problems of algorithms and AI through the lens of sociological theory will benefit from this timely book. Readers more familiar with algorithm systems but with limited understanding of sociological theory may also benefit from Machine Habitus. Most importantly, the book is a must-read text for any sociology of algorithms or AI undergraduate or graduate course.