Does Humanizing Chatbots Promote or Hinder Voice Shopping?
We rely on conversational agents (CAs) for obtaining a variety of information, but rarely for shopping. A primary design characteristic of CAs shaping user acceptance ever since their inception, is their anthropomorphic nature. They have become increasingly human-like, taking on human roles such as assistant, friend, or servant. While anthropomorphic features are often intended to foster engagement, voice shopping continues to lag behind market expectations. Could humanizing CAs undermine interaction quality rather than enhancing user experience through greater engagement? We investigated whether the effectiveness of humanized CAs depends on users’ relational perceptions, specifically whether they view the CA like a servant, friend, or rational agent. In an online experiment (N = 407), participants viewed standard or humanized videos of voice shopping interactions. Results showed that humanized design negatively affects perceived attitudes toward voice shopping. Furthermore, humanized design decreased perceived benefits when the CA was perceived more like a servant. Findings question widespread design emphasis on humanization and inspire new designs for CAs.