A Review of Explainable Artificial Intelligence in Manufacturing

AI in manufacturing delivers real benefits but creates a trust problem: While AI systems improve production efficiency, performance, and safety in manufacturing, their "black box" nature—where even developers cannot easily explain how they reach decisions—undermines human trust in these systems, which is critical when AI recommendations guide important operational choices.

Explainability is essential for responsible deployment: The paper identifies explainable AI (XAI) techniques as a necessary solution to increase transparency and make AI decision-making intelligible to humans, suggesting that opacity should not be accepted as an inevitable trade-off for accuracy.

Multiple evaluation methods exist for XAI techniques: The review describes various metrics to assess whether XAI methods actually succeed in making AI systems more understandable, indicating that organizations can measure and compare different approaches rather than relying on subjective judgments about transparency.

Manufacturing is a high-stakes domain requiring explainability: The paper's focus on manufacturing applications underscores that XAI matters most in contexts where AI failures have direct consequences for worker safety, product quality, and operational decisions—not just in low-risk applications.

XAI solutions are already being applied in practice: The paper describes "several application scenarios in the manufacturing domain," suggesting that explainable AI techniques are not theoretical concepts but are being implemented in real-world manufacturing settings today.