Asian Journal of Economics, Finance and Management

Artificial intelligence (AI)-enabled chatbots and conversational assistants have become prominent customer-facing interfaces in e-marketing, yet their contribution to customer experience depends on more than automation speed. This critical narrative review evaluates what is known about the effects of chatbots and AI assistants on customer experience, with Jordan as the focal context and international evidence used to interpret mechanisms and boundary conditions. Literature published from 2016 to 29 June 2026 was considered, with earlier foundational material included where conceptually necessary. The synthesis integrates Jordanian studies of banking, social-media marketing, manufacturing, small and medium-sized enterprises, consumer industries and digital public services with experimental and survey evidence on service quality, anthropomorphism, trust, privacy, disclosure, service failure and continuance. The evidence indicates that conversational AI can improve convenience, information access, perceived control, personalisation and satisfaction when systems are reliable, useful and capable of resolving routine tasks. Jordanian banking research particularly supports reliability, personalisation and customer empowerment as proximal determinants of satisfaction. However, responsiveness alone is not consistently sufficient, and human-like design is conditional rather than universally beneficial. Anthropomorphic cues may strengthen social presence and warmth in low-friction encounters, but can raise expectations and worsen evaluations when the system misunderstands an angry or high-stakes customer. Trust, privacy risk, transparency regarding machine identity, linguistic fit, escalation to human agents and integration with wider digital service quality therefore shape whether automation enhances or degrades the journey. The Jordanian evidence base remains dominated by cross-sectional self-report surveys, convenience sampling and partial least-squares modelling, with limited behavioural, longitudinal, experimental and Arabic-language interaction research. The review concludes that the strongest case for conversational AI in Jordanian e-marketing is a hybrid, task-contingent model that automates predictable interactions while preserving credible human escalation, transparent data practices and context-sensitive personalisation. Future work should prioritise field experiments, behavioural outcomes, Arabic dialect performance, vulnerable and less digitally confident consumers, and cross-sector longitudinal evaluation.