Enhancing the Product Quality of the Injection Process Using eXplainable Artificial Intelligence

XAI makes AI decisions understandable in manufacturing: The study uses explainable AI techniques (specifically SHAP analysis) to identify which specific factors in the injection molding process most influence product quality, moving beyond "black box" predictions to show why the AI makes its recommendations.

Transparency led to significant quality improvements: By understanding which process variables mattered most, manufacturers could optimize control ranges for those features, reducing defect rates from 1.00% to as low as 0.13%—demonstrating that explainability isn't just about understanding AI, but about achieving better real-world outcomes.

XAI enables human oversight in critical processes: Rather than replacing human decision-making, the explainable approach allows manufacturing engineers to validate the AI's reasoning, understand the optimal control parameters, and maintain meaningful human control over production systems that affect product quality.

Responsible AI implementation requires interpretable models: The researchers deliberately chose tree-based algorithms (XGBoost and LightGBM) specifically because they work well with explainability tools, suggesting that responsible AI development sometimes means selecting methods that prioritize interpretability alongside performance.