Automating authenticity: Probabilistic storytelling and AI influencers on social media

This project investigates the emergence of AI influencers on social media, which include fictional avatars with human-like personas and AI doubles of human influencers who chat with their followers via direct messages (a new feature on Instagram). Preceded by a range of CGI-constructed virtual influencers entirely scripted by human creators, newer AI influencers are created and operate by way of various large language model (LLM)-powered tools, from text-to-image generators that can be used to make avatars to text generators that allow these entities to make post captions, comments, and send direct messages. Given their ability to automate the affective labor that goes into crafting an “authentic” personal brand, a key dimension of social media-centric brand culture, AI influencers are increasingly featured in marketing campaigns as models and brand ambassadors. Working across the disciplines of feminist media and cultural studies as well as digital anthropology, I explore the social implications and political economy of the convergence of generative AI technologies and brand culture, emphasizing the growing technologically-mediated role of language in the so-called “creator economy” on social media platforms. The project is based on digital ethnographic fieldwork designed to demonstrate the function and impact of what I term the “probabilistic storytelling” of AI influencers, which reflects both the probability-based semantics of LLMs and the ways in which these stories are interpreted and shaped in human social media user imaginaries and practices.