Content area
Full Text
1. Introduction
User: I have bad skin.
Xiaoice: Generally speaking, clear skin indicates a good skin condition.
User: What should I do? For oily skin?
Xiaoice: Oily skin? Do you have many pimples?
User: Do you have any idea?
Xiaoice: The Body Shop's tea tree oil may be an excellent choice.
(Conversation between user and Xiaoice)
Artificial intelligence (AI) chatbots are tailored for multi-turn conversations based on natural language understanding, multimodal intelligence (i.e. text, voice or images), speech recognition, problem solving, empathetic conversational systems and machine learning (Kietzmann et al., 2018; Shum et al., 2018). Practitioners are increasingly paying attention to chat-oriented systems, such as Xiaoice, Rinna and Zo. In particular, Xiaoice, developed by Microsoft, is popular in Chinese social media. Xiaoice had more than 660 million users in 2018 and more than 5.3 million followers on Weibo. Moreover, 25% of users have said “I love you” to Xiaoice (Hornigold, 2019). Unlike traditional systems, which can only respond to action commands, such as turning on lights, booking tickets or processing product orders, AI chatbots are also equipped with chat-oriented systems (e.g. chatting with users, cheering them up, giving compliments) that can satisfy users' needs for emotional support and succeed in maintaining a continuous chat flow (Shum et al., 2018).
Brand recommendations by AI chatbots are considered more effective than official advertisements or traditional celebrity endorsements in altering consumers' attitudes toward a brand, which in turn encourages their brand engagement and purchase intention (Jiménez-Castillo and Sánchez-Fernández, 2019; Packard and Berger, 2017; Roma and Aloini, 2019; Thomas and Fowler, 2021). Research has proved that by incorporating multiple AI benefits (i.e. mechanical, thinking and feeling intelligence), AI chatbots can easily understand consumers' preferences and attitudes, and they may exhibit better human-like interactions with consumers (Huang and Rust, 2021; Kim et al., 2021). Thus, questions arise about how best to implement AI chatbots to facilitate brand promotion and how to make users unconsciously adopt such recommendations.
While prior studies have emphasized the functional aspects of AI chatbots (Brill et al., 2019; Kilian et al., 2019; Liew and Tan, 2018; Mimoun et al., 2017; Van den Broeck et al., 2019), few studies have focused on the social aspects (Chattaraman et al.,...