Determinants of AI-Enabled Quality Control Adoption Intention in Manufacturing SMEs: An Integrated TOE–TAM Analysis Using PLS-SEM, IPMA, and fsQCA

AI-enabled quality control (AI-QC) tools are increasingly available to manufacturing SMEs in emerging economies, yet the firm-level conditions associated with their adoption remain underexamined. Building on the Technology–Organization–Environment (TOE) framework of Tornatzky and Fleischer, integrated with the perceived usefulness and perceived ease-of-use constructs of the Technology Acceptance Model (TAM), this study examines the determinants of AI-QC adoption intention, and its association with operational performance improvement, in 284 manufacturing SMEs from Turkey, Malaysia, and Egypt. The focal dependent construct is adoption intention rather than realized adoption. The AI-QC technologies considered are machine learning defect detection, computer vision inspection, predictive maintenance, and digital twin integration. Three complementary analytical procedures are applied to the same data: partial least squares structural equation modeling (PLS-SEM) to estimate the strength of the modeled associations, importance–performance map analysis (IPMA) to identify high-importance but low-performance predictors, and fuzzy-set qualitative comparative analysis (fsQCA) to identify combinations of conditions jointly sufficient for high adoption intention. The PLS-SEM estimates indicate positive associations for the technological, organizational, and environmental predictors, with top management support, perceived usefulness, and organizational readiness showing the largest coefficients and data security concern showing a negative association; effect magnitudes varied considerably, and several were small. The IPMA results indicate that the two most important predictors exhibit comparatively low performance scores in the sample. The fsQCA results identify three configurations associated with high adoption intention. Because the design is cross-sectional and based on self-reported, single-respondent data, the findings are interpreted as associations rather than causal effects. The paper concludes with guidance for SME managers, AI technology vendors, and industrial policymakers.