Determinants of customer acceptance in AI-powered conversational agents: A systematic literature review and research agenda

Abstract

As conversational AI agents (CAIAs) become embedded in customer – firm interactions, the theoretical models employed to account for their acceptance increasingly appear inappropriate. This systematic literature review synthesizes 58 empirical studies (2018–2025) and exposes a core tension: established frameworks such as the Technology Acceptance Model (TAM) and the Unified Theory of Acceptance and Use of Technology (UTAUT) rest on assumptions of instrumentally rational users evaluating stable system attributes, yet CAIAs are adaptive, probabilistic, and socially responsive in ways that render those assumptions problematic. The resulting mismatch, we argue, is responsible for contradictory empirical findings, particularly at the intersection of trust, anthropomorphism, and perceived risk. Where rationalist models fall short, the Computers as Social Actors (CASA) paradigm proves more apt at capturing relational dynamics, though it too offers only a partial account. By distinguishing between utilitarian and relational agent types and between high-stakes and low-stakes decision contexts, the review uncovers boundary conditions that earlier syntheses have left unexamined. Drawing on consumer decision-making, value co-creation, and relationship management scholarship, we develop a marketing-centred research agenda comprising five theoretically grounded directions. Managerial guidance is offered for CAIA design attuned to distinct phases of the customer journey.