Social and psychological factors in AI chatbot trust-building

Full abstract

Read the full abstract

Purpose This study examines the aspects driving the formation of effective brand-human–AI connections. It explores the impact of consumers’ trust in AI-enabled e-commerce chatbots on their buying intention, e-loyalty (EL) and reuse intention (RUI). The study develops a conceptual model, drawing on social support theory, psychological reactance theory and the uncanny valley effect. Design/methodology/approach We collected data from 429 AI-enabled shopping chatbot users in Malaysia, employing a convenience sampling approach within a specific purposive frame, and analysed them using partial least squares structural equation modeling (PLS-SEM) via SmartPLS4. Findings Findings indicate that user autonomy with the chatbot and social role enhances consumers’ trust in chatbots. This trust, in turn, positively impacts customers’ purchase intentions, e-loyalty and chatbot RUI. However, the chatbot’s humanlikeness does not affect consumers’ trust in chatbots. The study also revealed that increasing a chatbot agent’s humanlike qualities increases users’ perception of eeriness, which decreases users’ trust in chatbots. Nevertheless, the chatbot’s humanlike attributes and user eeriness were significantly influenced by users’ familiarity with it. Originality/value This study’s theoretical and practical implications offer valuable insights for enhancing the online shopping experience through anthropomorphised chatbot agents in e-commerce.