The transition to clean energy demands efficient and scalable electrochemical energy storage (EES) systems, including metal-ion and metal–sulfur batteries, as well as supercapacitors. Carbon materials are essential for EES owing to their exceptional electrical conductivity, electrochemical stability, and structural versatility, but their complex structures hinder structure–property mappings, thereby limiting traditional trial-and-error design. Artificial intelligence (AI) and machine learning (ML) offer transformative solutions by uncovering hidden structure–property relations and accelerating material discovery. Although various EES materials have been benefited from these advances, the study of carbon materials using this new paradigm is in its early stages. This review provides a comprehensive evaluation of AI-assisted methods applied to carbon materials across diverse EES systems, including metal-ion and metal–sulfur batteries, as well as supercapacitors. We systematically summarize core ML workflows in materials science, spanning data acquisition, feature engineering, and predictive modeling while highlighting AI-driven theoretical simulations, such as ML interatomic potentials. Finally, we outline emerging paradigms and prospects, emphasizing the integration of large language models, active learning to address data scarcity, and the development of autonomous self-driving laboratories for next-generation carbon electrode design. This review also presents challenges and limitations of applying AI-assisted methods to research on carbon materials for EES.

As a carbon resource with abundant nitrogen, polyacrylonitrile (PAN) has been used for the key raw materials to produce carbon materials. However, the direct carbonization will lead to the cementation of PAN particles, which is adverse to the subsequent activation. In this work, a hybrid porous carbon (HPC) was synthesized via dry ball-milling of thermal-reduced graphene and polyacrylonitrile, and subsequent stabilization and KOH activation. The effect of the ratio of graphene and polyacrylonitrile, along with activation treatment on the properties of hybrid porous carbon are systematically studied. The results demonstrate that the existence of graphene nanosheets enable the fast dissipation of heat produced in ball-milling process and avoid the cementation of PAN particles, while PAN particles act as spacers to prevent graphene restacking. The as-obtained polyacrylonitrile/graphene precursor is a loose powder, which favors the dispersion of activation agents within the precursor and results in a more homogeneous and effective activation. Meanwhile, graphene works as a 3D micro-current collector, thus providing a conductive network for convenient charge transfer in HPC. Based on the electrochemical characterization in either water or a non-aqueous electrolyte, HPC exhibits a superior capacitive behavior because of its well-developed porosity, large specific surface area, excellent electrical conductivity as well as nitrogen/oxygen heteroatom induced pseudo-capacitance. Particularly, HPC-4 offers a high energy density of 30.38 W·h/kg at a power density of 337.5 W/kg in TEABF4/EC-DMC electrolyte. This work provides an easy-to-operate and efficient synthesis strategy for developing porous carbon electrode towards supercapacitors with high power and energy density.