Text me if you can! The influence of modality in consumer interactions with AI digital assistants: an abstract

AI-based digital assistants are Internet-enabled devices which provide daily technical, administrative and social assistance to their users and can be used as tools for online shopping, learning, controlling other smart applications and devices as well as for communications and companionship (Han & Yang, 2018; Santos et al., 2016). In particular, major tech companies are heavily investing in the development of AI digital assistant primarily designed to facilitate voice interactions between users and the systems (e.g., Amazon Alexa, Apple Siri, Google Assistant, Microsoft Cortana). However, many of the current voice-activated digital assistants offer additional modality options to interact with their users, including “type-in” features. For example, recent versions of Amazon Alexa come with a keyboard option that allows users to type in their requests or read Alexa’s replies. While the ways in which we interact with technology, and particularly with digital voice-assistants, is variegate, few studies have explored the influence that digital voice assistant modality has on consumers responses and behaviors (Melzner et al., 2020). The present research aims to fill this gap and, by focusing on consumers-digital assistants’ interactions, it investigates the effect of interaction modality (i.e., voice vs text) on consumers feelings of power, comfort and recommendation acceptance. Specifically, we posit that the modality through which consumers interact with the conversational digital assistant (text vs voice) may change consumers’ responses and impact their psychological states and subsequent behavior (Melumad et al., 2020). Since higher level of perceived anthropomorphism have been proved to be detrimental to consumers perceptions of power and control (Jörling et al., 2019), we expect that when individuals interact with AI digital assistants through voice (vs text) interfaces, their feelings of power over the technology as well as comfort decrease (increase) resulting in lower (higher) recommendation acceptance. Moreover, we investigate the moderating role of users’ previous experience with the digital assistant. Previous studies showed that when interacting with conversational agents, consumers may perceive the interaction differently depending on whether they already have experience of communicating with a digital assistant (Hildebrand et al., 2019). Building on this we suggest that previous experience moderates the effect of the digital assistant interaction modality on perceived power, comfort, and recommendation acceptance. Specifically, we expect the negative effect of voice-based interactions to be reversed with increasing level of prior experience. Preliminary findings from two experimental studies provide support to hour hypotheses and show that voice (vs text) interactions with AI conversational digital assistants decrease consumers’ willingness to accept recommendation through a decrease of comfort and power.