How to improve voice assistant evaluations: Understanding the role of attachment with a socio-technical systems perspective

Abstract

With the widespread adoption of artificial intelligence (AI) powered voice assistants (VAs) and increasingly homogeneous competition among enterprises, it is critically important to understand the driving factors behind consumer VA evaluations. Drawing on attachment theory and socio-technical systems theory, this study proposes a theoretical model to examine how social and technical attributes of VAs impact consumer evaluation behavior. To test the model, a two-wave longitudinal survey was conducted among 462 valid samples in China and analyzed using a multi-method approach, including partial least squares structural equation modeling (PLS-SEM) and fuzzy-set qualitative comparative analysis (fsQCA). PLS-SEM findings reveal that consumers' VA evaluations (continuance and word-of-mouth intentions) are primarily influenced by individuals' emotional and functional attachments toward VAs. These attachments, in turn, are determined by social attributes (interactivity, natural speech, and design aesthetics) and technical attributes (accuracy, connectivity, and personalization). Furthermore, the results indicate that social and technology anxiety play a moderating role in the relationship between VA attributes and attachments. The fsQCA analysis supports the PLS-SEM findings and identifies three configuration paths for high continuance intention and three configuration paths for positive word-of-mouth intention toward VAs. These findings provide novel insights into both theory and practice.