Leveraging cubism principles to foster innovation in manufacturing systems: an AI-enabled mixed-methods analysis

Purpose This study examines how Cubist principles – modularity, abstraction and multiple perspectives – relate to innovation in manufacturing systems. It identifies factors associated with innovation, organizational challenges and approaches for strengthening innovation capabilities. Design/methodology/approach A mixed-methods design combined quantitative and qualitative analyses with machine-learning techniques. Data were collected through a structured questionnaire comprising closed- and open-ended questions from employees across manufacturing roles. Quantitative analysis included demographic analysis, confirmatory factor analysis (CFA), structural equation modeling (SEM) and machine-learning models, including neural networks and random forests. Qualitative analysis involved text preprocessing, word-frequency analysis, sentiment analysis and latent dirichlet allocation (LDA) topic modeling. The results were integrated to provide a comprehensive understanding of innovation in manufacturing systems. Findings Modularity, abstraction and multiple perspectives were significantly associated with innovation, with multiple perspectives showing the strongest association (path coefficient = 0.52). Random forest and neural network models demonstrated strong predictive performance (R2 > 0.80). Qualitative findings highlighted resistance to change and resource constraints as key challenges, while advanced tools and cross-departmental collaboration emerged as potential solutions. Sentiment analysis showed a predominantly positive outlook (56.25% positive responses). Research limitations/implications Manufacturing organizations may strengthen innovation by promoting modular design, investing in AI and 3D modeling, fostering cross-departmental collaboration, addressing resource constraints and resistance to change and providing employee training. Originality/value This study contributes by integrating Cubist principles with manufacturing innovation through a mixed-methods approach that combines statistical rigor, machine-learning prediction and qualitative insights.