Alexa is My Smart Friend: Children’s Cognitive, Social, and Moral Attributions Towards Voice-Assistants

• In Event: Poster Session 01
  In Poster Session: PS 01 Topic - Technology, Media & Child Development

Abstract

As voice-assistants such as Alexa, Siri, and Google become increasingly present in homes and classrooms, children are interacting with these devices more frequently (Common Sense Media, 2019; Lovato, Piper, & Wartella, 2019). Children are willing to treat humanoid robots as intelligent, autonomous, and social entities (Kahn et al., 2012); however, it is unknown if children treat voice-assistants similarly. This study looks at whether children ascribe human-like cognitive, social, and moral attributes to voice-assistants. Seventy-four children ages 6 to 10 (Mage=8.12; 33 females) were introduced to three voice-assistant speakers: Alexa, Siri, and Google. Children chose the device they found most familiar and answered nine questions about its characteristics. The questions focused on three types of attributes: cognitive, moral, and social (see Table 1). Children were also asked to justify their answers to three questions (one from each category; see Table 2). Two coders classified the children’s responses into 6 categories (see Table 2) with 96% interrater agreement. Correlational analyses explored how children’s age related to the total number of attributions they made to the voice-assistant within each question category. Younger children were more likely to attribute social (r(74)=-.30, p=.008) and moral (r(74)=-.27, p=.017) characteristics to the voice-assistant than older children; however, cognitive attributions were not correlated with age. When justifying their responses to the question of whether the voice-assistant could be interested in science, Children who gave affirmative responses (47% of the sample) frequently indicated that voice-assistants have cognitive capabilities, such as thinking, learning, and preferences, and children who gave negative responses (53%) indicated that voice-assistants do not have cognitive capabilities. Additionally, children who did not believe the device could be interested in science were more likely to describe the voice-assistant as an artifact. Children provided conventional justifications, especially related to the device’s ability or inability to keep information private, as the primary reason for both agreeing (43%) and disagreeing (57%) that the device could keep a secret. Finally, in explaining whether it is okay to sell a voice-assistant (which 80% of the sample affirmed), the majority of children cited social conventions regarding ownership of property. Taken together, these results indicate that children’s attributions towards familiar voice-assistants change with age. Younger children may not fully understand the capacities of voice-assistants and thus be more likely to view them as human-like. However, with increasing age, children ascribe fewer social and moral qualities to voice-assistants. It may be that older children have more experience using these devices, and thus have a greater understanding of their capacities and limitations. Additionally, children who indicated that voice-assistants had cognitive, social, or moral attributes frequently described them in human-like terms, whereas children who did not attribute these qualities to voice-assistants were more likely to describe them as inanimate objects.

Authors

• Hailey Marie Streble, University of Louisville
  Presenting Author

• Lauren Girouard, University of Louisville
  Non-Presenting Author

• Judith H Danovitch, University of Louisville
  Non-Presenting Author