Recasting Service Quality for AI-Based Service

Abstract

Artificial intelligence service agents (AISA), such as chatbots and virtual assistants, are becoming increasingly pervasive in service. Research to date has not adequately addressed how the unique nature of AISA shape consumers’ service quality expectations. A deeper understanding of AISA service quality is important for their successful deployment in the service sector. To address this gap, we reviewed marketing and information systems literatures and conducted qualitative in-depth interviews with 37 informants, inclusive of 28 AISA users and nine AISA experts. We developed a conceptual framework for how consumers use and evaluate AISA. Twelve service quality dimensions emerged from the qualitative evidence representing AISA service quality, two of which align with AISA’s unique characteristics. The study extends the service quality theory to a new context and offers fresh insights for theory and practice. It culminates with a research agenda to advance research on AISA service quality.

Developing a service quality scale for artificial intelligence service agents

“…The second stage aimed to both validate identified dimensions and identify new dimensions that might be applicable to AISA service quality and included users of chatbots and virtual assistants, as common, popular types of AISA in the high (humanlike) touch – high-tech paradigm. The 12 dimensions identified by Noor et al (2021b) are summarized in Table 2.…”

“…It is a global assessment and attitude consisting of consumer judgements of AISA service performance. We adopt the dimensions proposed by Noor et al (2021b) for which we identify measure items at the perceptual level to effectively capture the abstract nature of service quality comparisons which consumers make across categories (Zeithaml, 1988). These service quality perceptions can be formed through AISA usage and in turn affect various outcomes such as customer satisfaction and loyalty (Cronin et al , 2000).…”

“…Following Noor et al (2021b), we also use chatbots and virtual assistants in this study as common forms of AISA used by service providers. We argue that this choice is appropriate due to the popularity and wide availability of these types of AISA, which is important in facilitating the recruitment of participants with experience in the use of these types of AISA, therefore providing access to meaningful responses for the subsequent stages of scale development and validation.…”

“…Further, based on Noor et al ’s(2021b) qualitative evidence, additional measure items were introduced to better capture the intended dimensions (Netemeyer et al , 2003). For example, one of the items “the AISA uses its own ‘judgment’ to complete a task” was added from Noor et al ’s(2021b) findings. After screening for irrelevant, redundant, ambiguous and double-barrelled statements, an initial battery of 85 items was produced.…”

“…We extend their work in this paper. Specifically, following on a strong tradition of established, proven scale-development processes and techniques (Churchill, 1979; Netemeyer et al , 2003), we build on Noor et al ’s (2021b) proposed qualitative dimensions to construct, refine and validate a multiple-item scale for measuring the AISA service quality of chatbots and virtual assistants, two types of AI service agents that are commonly used by service providers in the current market.…”