Investigating gender-stereotyped interactions with virtual agents in public spaces

Current research on the impact of gender appearance in virtual agents and social robots highlights the danger of transmitting and solidifying existing gender stereotypes. To investigate gender-stereotyped interaction at public spaces in dependency of virtual agents gender, we varied the gender of a virtual agent at a metro station. We used an ethnographic study approach, combining a two-day behavior observation with semi-structured interviews with descriptive and qualitative system log analysis of four weeks. Our results show that topics of conversation differ in dependency of the virtual agents gender: the male virtual agent was asked about topics such as brothels, drugs and alcohol and insulted frequently, while the female one was asked for relationship status or about flirting.