Looking at chatbots in the service sector, there is a clear trend towards anthropomorphized, often feminized avatars, which is currently accompanied by a lot of criticism. This study investigated users? preferences for chatbot avatars (in terms of visual human likeness and gender) and how different looks of chatbots influence intentions to use them. 1,064 participants in an online experiment were assigned to a functional prototype of a customer service chatbot for banking, either anthropomorphic or non-anthropomorphic in appearance. After interacting with the chatbot, e.g. requesting a bank?s opening hours, participants indicated its perceived gender, reported how much they would like to use it, and chose from a range of design options what they would ideally want the chatbot to look like (abstract icons, humanlike avatars in male, female or gender-ambiguous form). Findings showed that male users had a significantly larger preference for a female-looking chatbot than female users, who were more open to gender-ambiguous designs. Additionally, participants who perceived the gender of the assigned chatbot to match their personal preference had a significantly higher intention to use it. Based on the results, important implications for chatbot design in the light of gender roles and stereotypes can be discussed. By including a gender-ambiguous avatar in the study, we investigated the potential of non-gendered technology for the future.
Forschungs-