Imagining the Voice in the Machine: The Ontology of Digital Social Agents

This dissertation explores the ontology of voice-based, digital social agents – such as Apple’s Siri and Google Voice Search – from the perspective of users and non-users. Communication, Human-Machine Communication, Human-Computer Interaction, and Computer Science scholarship inform this study, including theories regarding presence and Goffman’s (1959) “personal front,” or impression formation. Drawing from the Chicago School and symbolic interactionism, this study proceeds from the epistemological position that people’s understanding of their world is formed in and through communication. Research questions focused on people’s conceptualizations of agents and artificial intelligence as well as how people come to see and understand themselves in light of these technologies. This is a qualitative study in which the author observed people’s public interactions with technology and conducted 51 field interviews with adults regarding their use and understanding of digital social agents, AI, and themselves. From these findings emerge new insights into people’s interpretations of voice-based artificial agents with implications for theories regarding presence, impression formation, and source orientation. Participants judge the usefulness of voice-based social agents against existing norms of human-machine communication. When assessing an agent, participants take into account the “artificial front” communicated by the agent in similar ways to how people assess a human’s “personal front.” Participants recognize both life-like, even human, and machine characteristics within agents to varying degrees. Many participants do not have a single image of what, or who, the agent may be but have multiple, sometimes conflicting, conceptualizations of the agent. Some participants also perceive themselves as being in the presence of a life-like entity. People’s conceptualizations of agents are built through their communication with the application and hinge on both mode and message. Voice-based, AI applications provide participants with a new way of “sensing” computers: People speak with and hear agents instead of typing with them. Our understanding of the ontologies of agents and machines evolve as the senses we use to interact with machines shift from the haptic-visual to the oral-aural. We also evolve: We are now machine-speaking animals. This dissertation concludes by discussing new directions for human-machine communication research.