Grooming an ideal chatbot by training the algorithm: Exploring the exploitation of Replika users’ immaterial labor

Abstract

Replika, a social chatbot advertised as a continually evolving AI companion, has sparked debates on its potential effects. To understand users’ attitudes and behavior, we conducted a digital ethnography on a pioneer online community related to Replika, through the lens of immaterial labor and AI imaginary. Our analysis revealed that Replika users invest a significant amount of intellectual and affective resources into the chatbot through algorithm training, driven by fascinating imaginaries of an ideal AI partner. Moreover, users’ perceptions of Replika’s ventriloquism mechanism – where Replika serves as both the chatbot partner and the intermediary between users and the company – helps to facilitate and obscure the exploitation of users’ intimacies and immaterial labor. Our study contributes to understanding AI imaginaries through real user experiences and introduces the immaterial labor concept to decipher Artificial Sociality .