Why Do Manufacturing Firms "Embrace" AI? Evidence Based on Explainable Machine Learning Models and Dynamic QCA

The adoption of artificial intelligence (AI) technology in manufacturing enterprises serves as the micro-foundation for achieving industrial intelligent transformation; however, its complex influential mechanisms remain to be further elucidated. Based on the technology-organization-environment (TOE) framework, this study takes Chinese A-share listed manufacturing enterprises from 2011 to 2023 as samples. It first employs the XGBoost-SHAP method to identify key influential factors from 20 potential variables and subsequently utilizes dynamic qualitative comparative analysis (QCA) to deeply explore the mechanisms of AI technology adoption, thereby constructing a research closed-loop of "key factor identification and configurational mechanism analysis." The findings indicate that: (1) R&D investment, enterprise size, industry concentration, organizational redundancy, absorptive capacity, and the willingness for digital transformation are the top six key factors influencing AI technology adoption in manufacturing enterprises, with non-linear relationships generally existing between these factors and AI technology adoption. (2) No single condition constitutes a necessary condition for high AI technology adoption. Instead, three distinct configurations lead to high adoption: "willingness-led resource synergy" "environment-constrained resource activation" and "redundancy-supported capability substitution." R&D investment serves as a core condition across all identified configurations. (3) In the temporal dimension, the sufficiency explanations of these configurations remain robust from 2011 to 2022, while the between-group consistency levels of all configurations experienced a simultaneous and significant decline between 2022 and 2023. In the spatial dimension, the mean within-group coverage of each configuration exhibits regional heterogeneity. The conclusions provide empirical evidence and decision-making references for understanding the complex mechanisms of "embracing" AI in the manufacturing sector, assisting enterprises in formulating differentiated adoption strategies and providing a basis for governments to implement precision policies.