Abstract: Chemical mechanical polishing (CMP) is a key technology for semiconductor wafer planarization, ultraprecision machining of high-end optical components, and low-damage processing of hard and brittle materials. It plays a critical role in nanoscale surface quality control and atomic-scale surface fabrication. CMP involves multiple interacting physical and chemical phenomena, including abrasive particle motion, interfacial contact, slurry transport, and surface chemical reactions. Complex nonlinear relationships exist among process parameters, process states, material removal, surface quality, defect formation, and process stability. Conventional empirical models and trial-and-error approaches therefore struggle to meet the increasing demands for manufacturing precision, stability, and consistency. In recent years, artificial intelligence methods have been increasingly applied to CMP, including machine learning, deep learning, reinforcement learning, and hybrid physics- and data-driven modeling. Their applications cover mechanism modeling, process monitoring, performance prediction, and process control. Recent advances are reviewed in four areas material removal and damage modeling, process monitoring and state assessment, machining performance prediction and quality evaluation, and process optimization and closed-loop control. Particular attention is given to several key challenges, including data quality, model interpretability, generalization across operating conditions, and sensing-prediction-control integration. Future research directions are also discussed, with an emphasis on physics-informed modeling, multisource data fusion, and reliable closed-loop control. This review provides a reference for the intelligent development and engineering application of CMP technology.