This work presents the state-of-the-art AI adoption for manufacturing optimization in European research projects, considering the three levels: system level, workstation level, and process level. As an example of a good practice for the process level, metaheuristics, i.e., evolutionary algorithms, are broadly discussed in terms of their employment for optimizing processes. The main findings of a recent literature review are short listed, revealing the most frequently and the most successfully applied algorithms for addressing process optimization. The evolutionary algorithms-based optimization methodology is briefly illustrated, along with its implementation on multiple case studies, employing various metaheuristics. The performances of different algorithms are carefully analyzed, demonstrating superior outcomes of the Jaya algorithm and Teaching Learning-Based Optimization (TLBO) algorithm, followed by the Particle Swarm Optimization (PSO) algorithm. Finally, recent developments and perspectives of Generative AI (GenAI) application in a manufacturing context are introduced, identifying the main issues to be addressed to ensure wider adoption of GenAI in a manufacturing ecosystem.