Customer acceptance and objection of mobile AR shopping apps: a dual-path analysis

Purpose This study investigates the drivers of customer acceptance and objection towards “augmented reality” (AR) shopping applications. Current technology adoption models provide limited insight into consumers’ simultaneous attraction to and resistance towards AR. To address this gap, this study adapts the AI device use acceptance (AIDUA) framework to the AR shopping context by replacing anthropomorphism with four antecedents, including perceived intelligence, aesthetic quality, perceived control and novelty value. Privacy risk and social innovativeness are incorporated as boundary conditions to capture when positive evaluations lead to acceptance and when they lead to objection. Design/methodology/approach Survey data from 539 consumers were analysed using a dual-stage methodology combining partial least squares structural equation modelling (PLS-SEM) and artificial neural networks Findings The dual-stage analysis, combining PLS-SEM and artificial neural network methods, demonstrates that both technological and experiential attributes significantly impact attitudes and intentions towards AR shopping apps. Results show that social influence, hedonic motivation, perceived intelligence, performance expectancy and effort expectancy are pivotal for AR acceptance, while privacy risk to personal privacy and social innovativeness play key moderating roles. Practical implications The findings provide guidance for developers and retailers on designing mobile AR shopping applications that emphasize experiential quality while addressing consumer privacy concerns. Marketers can leverage social influence and novelty features to increase adoption and minimize objections. Originality/value This study provides several contributions to AR adoption research. First, by modelling acceptance and objection as coexisting pathways, it moves beyond single-continuum frameworks such as TAM and UTAUT and offers a more comprehensive account of the interest–use gap in AR shopping. Second, it adapts AIDUA’s primary appraisal layer to a non-humanoid setting by replacing anthropomorphism with four AR-specific antecedents, extending the model’s applicability beyond service robots and humanoid agents. Third, it incorporates privacy risk and social innovativeness as boundary conditions, clarifying when and for whom acceptance and objection are strengthened.