We are entering an era of AI-Mediated Communication (AI-MC) where interpersonal communication is not only mediated by technology, but is optimized, augmented, or generated by artificial intelligence. Our study takes a first look at the potential impact of AI-MC on online self-presentation. In three experiments we test whether people find Airbnb hosts less trustworthy if they believe their profiles have been written by AI. We observe a new phenomenon that we term the Replicant Effect: Only when participants thought they saw a mixed set of AI- and human-written profiles, they mistrusted hosts whose profiles were labeled as or suspected to be written by AI. Our findings have implications for the design of systems that involve AI technologies in online self-presentation and chart a direction for future work that may upend or augment key aspects of Computer-Mediated Communication theory.

Self-presentation is a process that is significantly complicated by the rise of algorithmic social media feeds, which obscure information about one's audience and environment. User understandings of these systems, and therefore user ability to adapt to them, are limited, and have recently been explored through the lens of folk theories. To date, little is understood of how these theories are formed, and how they tie to the self-presentation process in social media. This paper presents an exploratory look at the folk theory formation process and the interplay between folk theories and self-presentation via a 28-participant interview study. Results suggest that people draw from diverse sources of information when forming folk theories, and that folk theories are more complex, multifaceted and malleable than previously assumed. This highlights the need to integrate folk theories into both social media systems and theories of self-presentation.

The ubiquity of mobile devices has resulted in more opportunities to interact with more people than ever before. Given a finite capacity for interaction with others, people commonly manage their availability by limiting others' access to them. Prior work has demonstrated the importance of doing so in a relationally sensitive way and identified the butler lie, in which deception is used to manage availability, as a common linguistic strategy. Two key limitations of existing exploratory work, however, are limited samples of primarily students and a focus on media properties in understanding ambiguity that enables butler lies to be plausible. This paper aims to address these issues via a broad field study of deception and butler lies using a novel message-sampling method employed via a custom mobile app. Results show clear evidence of butler lies occurring in a broader population, with some gender differences; and urge adoption of a multi-level framework for understanding ambiguity that also includes private information and infrastructure-level attributes of interaction media.