Theoretical Challenges of Human-Machine Interaction

Abstract

Abstract Machine learning systems and their interactions with humans challenge core assumptions of social theory. They are not completely defined by designers’ or users’ scripts. They hold an element of surprise and unpredictability. As a result, machines, similarly to humans, are increasingly able to produce contingency in their behavior. However, they achieve this through advanced techniques of information processing and pattern recognition, which are different from how humans relate to the world. Most notably, machines lack hermeneutic understanding. Despite this difference, they are becoming intelligible to humans as participants in communication—and vice versa. Sociology thus has to account for how humans and machines are able to interact with one another and recognize each other as intelligible agents despite operating in radically different ways. In practice, this is possible because of a specific kind of relation: the interface, which interconnects human and machine by keeping them apart. This chapter argues that a sociology of human-machine interaction requires a theory of interfaces. In response to these challenges and by drawing on an emerging interdisciplinary body of literature on interfaces, the chapter concludes by outlining three directions in which to study human-machine interaction at, through, and across the interface.