Gender Biases: The Interaction Between Children and Voice Assistants

Nowadays, children can access and interact with voice assistants (VAs) easily since the machines are more available in the family context. However, there is a gap in understanding how children conceptualize VAs when interacting with the machine. Specifically, this study aimed to examine whether children understand VAs as inanimate. Additionally, the study also compared how children and adults rate the VAs’ level of warmth and competence and the way they associate AI voices with feminine or masculine faces. Another goal was to explore how children and adults match the gender and competence/warmth of the VA with facial masculinity and femininity. The study employs the Computer as Social Actors (CASA) paradigm, where participants interact with a VA with different gender and competence/warmth conditions, while also using a Wizard of Oz design to simulate an autonomous VA. The study was a between-subjects design with children from 7- to -10 years old (n = 34) and adults from 18- to -35 years old (n = 61). The results indicated that children and adults rated the VA high on both warmth and competence, which are theoretically distinct dimensions. Interestingly, there were age differences in how people understood the VA through drawing. Adults were more likely to draw and describe the VA with human-like characteristics and children were more likely to draw and focus on the mechanical characteristics. This finding will help us design more suitable and beneficial AI systems for children’s learning by tailoring voice assistant interactions that align with children’s developmental understanding of technology. By recognizing how children conceptualize VAs in terms of warmth, competence, and gender, developers can create voice assistants that foster more effective, engaging, and age-appropriate educational experiences for children.