Toolkit or Omnipotent Friend? The role of Virtual Assistant for Adolescents and Young Adults and its Social Effects



With the increasing prevalence of AI, virtual assistants are engaging more with human life by providing services and information. Retale’s report showed that millennials are the biggest group of early chatbots adopters. To date, numerous experiments demonstrated that individuals’ social behaviors are different between human-human interactions and human-machine interactions (Nass, Moon & Green, 1997; Nass & Moon, 2000; Ho, Hancock & Miner, 2018), yet relatively few studies have investigated the attitude of adolescents/young adults when interacting with virtual assistants and social effects of Artificial Intelligence-Mediated Communication(AI-MC) (Hancock, Naaman & Levy, 2020). This paper attempts to address this limitation. The study reveals the findings of interviews with 14 users of virtual assistants to understand their interactional behaviors and interpersonal communication effects.

Virtual assistants, like Siri and Microsoft XiaoIce, put quite a few efforts on emotional recognition which are conducive to establish connections (Shum, He & Li, 2018). Humanlike cues are important factors to trigger social responses, as individuals might apply social rules and take computers as social actors (Reeves & Nass, 1996; Nass & Moon, 2000).

Time spent with any media is regarded to choose different channels of socialization (Larson & Kubey, 1983). According to Erickson’s theory of human development, adolescence is the final stage of basic socialization. Spending enough time with families and friends would facilitate teens’ formal adult socialization (Larson, 1983). However, the diversity of adolescents’ uses of media would cause conflict with other socializers (Arnett, 1995). The emergence of virtual assistants is potentially replacing other playmates, which we are unaware of its influences.

RQ1: How do adolescents and young adults evaluate the role of virtual assistants when socially interacting with it?

RQ2: Whether the AI-MC will exert a potential negative influence on interpersonal social interaction and socialization?


The data were collected during December 2019 and January 2020 by conducting in-depth, semi-structured interviews with 14 users of virtual assistants. Each interview lasted between 46 minutes and 94 minutes, with an average length of 61.33 minutes (SD=14.20). Participants are comprised of adolescents (10-17 years-old, n=7) and young adults (18-35 years-old, n=7). Interviews occurred either face-to-face (n=4) or via WeChat audio (n=10), depending on the location of participants. All the audio recordings of interviews were translated into transcripts, importing into the qualitative analysis software Atlas.ti for tracking open, axial, and selective coding, based on grounded theory (Glaser & Strauss, 2017).


In light of our findings, adolescents are more likely to treat virtual assistants as friends, while young adults take them as tools, less likely to establish intimate relationships. Compared to young adults, adolescents tend to show intimate disclosure behaviors to virtual assistants which are conducive to improve relational outcomes (Afifi & Guerrero, 2000; Ho et al, 2018). Additionally, when it refers to the potential negative effects on the socialization and interpersonal interaction, the third-person effects (Davison, 1983) are emerging. Especially for adolescents, they tend to believe that addicting to rely on virtual assistants would have a negative influence on their friends rather than themselves.