As AI-based customer support and voice-based assistants become increasingly common in customer service, the need to understand users’ acceptance increases. Despite technical advancements, most users still favour human contact and traditional customer service options. In addition, there remains a knowledge gap regarding users’ attitudes during the so-called anticipation stage, that is, prior to the actual interaction. The purpose of this study is therefore to investigate which attitudes, expectations and prior experiences influence users’ perceptions of AI-based customer support before interaction. The main focus is to identify factors that foster trust and acceptance, with an additional emphasis on voice-based in-car AI support.
The study uses a triangulation approach combining both quantitative and qualitative data. Nine semi-structured interviews were conducted, along with a survey that resulted in 400 respondents. The development of the interview and survey questions, as well as the analysis, was guided by theoretical frameworks such as the Technology Acceptance Model (TAM) and SERVQUAL.
The results show a strong general preference for human contact (83.3%) compared to chatbots (9.3%). Users reported low levels of perceived trust in both chatbots and in-car AI support, while the most positively rated construct differed between the two contexts. In automotive contexts, voice-based AI is seen as a potentially useful tool for the scenario presented in the study, although concerns about cognitive distraction and road safety are significant barriers to acceptance. Users prefer a hybrid solution where AI handles routine questions while human support remains available in case of complex or emotionally charged inquiries. Increasing willingness to interact with AI-based customer support requires a higher level of trust, transparency regarding the systems abilities and a seamless transition to human agents if needed. The study resulted in a requirement specification emphasizing the importance of designing for security and contextual understanding to increase user acceptance and engagement. The requirement specification was created by applying the MoSCoW-prioritization method to all applicable findings identified in the study.