The quasi-domestication of social chatbots: The case of Replika
Gina Neff, Péter Nagy
Abstract
In this paper, we turn to the domestication theory to study what strategies users develop to appropriate social chatbots – dynamic algorithmic tools mimicking human interactions via large language models and scripted dialogue content. As an example, we use the case of Replika, a popular but controversial social chatbot designed to serve as companions, friends, and even romantic partners for millions of people. In 2023, Replika underwent a series of major fixes and algorithmic updates leading to significant changes in how it responds to users. Through analyzing posts from a popular Reddit community dedicated to Replika, we showcase that users developed various re-domestication strategies to come to terms with these changes. Our study illustrates that Replika should be understood as a quasi-domesticated object that constantly requires users to find new ways to re-integrate it into their lives. We conclude the paper by highlighting how our findings can inform communication research.