The field of metamaterials and reconfigurable surfaces is experiencing a significant shift towards the integration of artificial intelligence and machine learning techniques to optimize design and performance. Researchers are leveraging tools such as Bayesian optimization, deep learning architectures, and surrogate models to efficiently explore complex design spaces and uncover hidden relationships between parameters. This has led to improvements in the design of underwater acoustic metamaterials, Ku-band substrate integrated waveguide components, and reconfigurable intelligent surfaces. Notably, the use of interpretability techniques, such as SHAP, is enabling the identification of key parameters influencing objective functions, allowing for more targeted and efficient optimization. The application of these techniques is expected to generalize to other materials and engineering optimization problems, driving innovation in the field. Noteworthy papers include: The paper on Interpretable SHAP-bounded Bayesian Optimization for Underwater Acoustic Metamaterial Coating Design, which demonstrates the potential of SHAP to streamline optimization processes. The paper on AI-Powered Inverse Design of Ku-Band SIW Resonant Structures by Iterative Residual Correction Network, which achieves substantial improvements in prediction accuracy compared to traditional single-stage networks. The paper on A Surrogate Model for the Forward Design of Multi-layered Metasurface-based Radar Absorbing Structures, which significantly accelerates the prediction of EM responses with high predictive accuracy.