In high-stakes manufacturing environments, the opacity of "black-box" Machine Learning (ML) models presents significant risks to safety, trust, and efficiency. While Explainable AI (XAI) offers a solution, its practical application landscape remains fragmented. This paper addresses this gap by providing a systematic review and novel synthesis of XAI adoption in manufacturing. Rather than merely summarizing known XAI techniques, we introduce a unique, application-centric categorization, analyzing 91 carefully selected articles across six key industrial domains: Predictive Maintenance, Additive Manufacturing, Process Optimization, Product Quality, Product Development, and Supply Chain Management. Our analysis reveals data-driven insights into publication trends and methodology choices, highlighting a definitive shift toward local, model-agnostic methods like SHAP. By structuring the review around manufacturing-specific challenges, we move beyond a general overview to provide a targeted analysis of the field’s current state. We conclude by identifying critical, domain-specific gaps and proposing actionable directions for future research to foster a more effective and trustworthy integration of advanced AI into the human workforce.

Apart from being an economic struggle, migration is first of all a societal challenge; most migrants come from different cultural and social contexts, do not speak the language of the host country, and are not familiar with its societal, administrative, and labour market infrastructure. This leaves them in need of dedicated personal assistance during their reception and integration. However, due to the continuously high number of people in need of attendance, public administrations and non-governmental organizations are often overstrained by this task. The objective of the Welcome Platform is to address the most pressing needs of migrants. The Platform incorporates advanced Embodied Conversational Agent and Virtual Reality technologies to support migrants in the context of reception, integration, and social inclusion in the host country. It has been successfully evaluated in trials with migrants in three European countries in view of potentially deviating needs at the municipal, regional, and national levels, respectively: the City of Hamm in Germany, Catalonia in Spain, and Greece. The results show that intelligent technologies can be a valuable supplementary tool for reducing the workload of personnel involved in migrant reception, integration, and inclusion.

In the realm of industrial energy management, the steel industry stands as a significant consumer, necessitating innovative approaches to optimize energy usage. This paper presents a comprehensive study leveraging machine learning models to analyze energy consumption data from DAEWOO Steel Co. Ltd in Gwangyang, South Korea. Utilizing datasets encompassing daily, monthly, and annual electricity usage, this research deploys various Naive Bayes models (Gaussian, Multi-nomial’ Complement, Bernoulli) to predict and analyze energy consumption patterns. Key to this study is the use of Permutation Feature Importance (PFI) and SHapley Additive exPlanations (SHAP) for feature importance analysis. With a particular focus on an advanced extension of SHAP that enhances the interpretability of model predictions by introducing a weighting factor based on feature importance derived from the model itself, Weighted SHapley Additive exPlanations (Weighted SHAP). This extension allows for a more nuanced assessment of feature contributions to the energy consumption patterns. By aligning with such advanced data exchange protocols, the comparative analysis of these models, enriched with the sophisticated SHAP methodology, provides valuable insights, contributing to a more efficient and sustainable energy management strategy in the steel industry. This alignment ensures that the energy management solutions proposed can be seamlessly integrated, furthering the advancements in smart manufacturing and Industry 4.0.

The proliferation of complex machine learning models, particularly deep neural networks (LeCun, Bengio, and Hinton 2015) and large ensemble methods such as gradient boosting (Chen and Guestrin 2016...