XAI-KG: knowledge graph to support XAI and decision-making in manufacturing

Explanations matter for trust in AI decisions: The paper emphasizes that understanding why an AI model makes a prediction is crucial for decision-makers to determine whether they can trust the forecast, particularly in manufacturing contexts like demand forecasting where business decisions have real consequences.

Current XAI methods lack quality assessment: While many techniques exist to explain AI model predictions, the researchers found that "little research was done on assessing the quality of those explanations and their influence on decision-making"—highlighting a gap between having explanations and having good explanations that actually help people make better decisions.

Feedback loops can improve AI systems: The proposed knowledge graph collects structured feedback on forecasts, explanations, and resulting user actions, creating a feedback mechanism that can simultaneously improve the underlying forecasting models, the quality of explanations provided, and the decision-making recommendations offered.

Domain-specific approaches are needed: Rather than one-size-fits-all XAI solutions, the researchers tailored their framework specifically for demand forecasting in manufacturing and validated it on real-world data, suggesting that responsible AI implementation requires understanding the particular context and stakes of each application.

Explainability supports human oversight: By connecting explanations to actual user decisions and outcomes, this approach enables organizations to monitor whether AI explanations are genuinely supporting human judgment or potentially misleading decision-makers.