DERIVING TRUST-SUPPORTING DESIGN KNOWLEDGE FOR AI-BASED CHATBOTS IN CUSTOMER SERVICE: A USE CASE FROM THE AUTOMOTIVE INDUSTRY

Abstract

In the automotive industry, companies are increasingly implementing Artificial Intelligence (AI)-based chatbots to support various processes, especially in the context of customer service. However, there currently is a lack of knowledge, especially systematically derived design knowledge, regarding customer trust in interacting with AI-based chatbots. In this context, a lack of security and transparency, limited social features, and the communication style and quality-related issues of AI-based chatbots are just a few aspects that inhibit customer trust in interacting with this innovative technology, thereby hindering the adoption of chatbots. To address this knowledge gap, we adopted a design theory-based approach and developed a design concept for trust-supporting design knowledge regarding customer interaction with an AI-based chatbot. Design science provides a structured development and evaluation process to support, for example, the adoption of AI-based chatbots. Drawing on trust-based literature, a use case in customer service in the automotive industry, and seven semi-structured expert interviews, we propose 10 meta/user requirements and four design principles for trust-supporting design elements as (e.g. social) signals (stimuli) regarding the interaction with AI-based chatbots. We developed two click prototypes over two evaluation cycles. Each evaluation included an online survey with 180 participants. The findings that were obtained make a valuable contribution to solving the described lack of design knowledge by developing and evaluating different design approaches in the form of prototypical user interfaces. Moreover, the results show that visible design elements such as transparent and factual security signals (stimuli) and trust seals have a significant impact on customer trust. © 2023 Taylor & Francis Group, LLC.