Proton exchange membrane (PEM) fuel cells play a pivotal role in a sustainable society through the direct conversion of hydrogen energy to electricity. Porous electrode materials, including porous media flow fields, gas diffusion layers, microporous layers, and catalyst layers, are essential for fuel cell operation, efficiency, and durability, in which complex multiphysics transport (e.g., hydrogen/oxygen transport, electron/proton conduction, heat transfer, and liquid water flow) and electrochemical reactions (e.g., the oxygen reduction reaction at the cathode and the hydrogen oxidation reaction at the anode) occur, as revealed by both experiments and multiphysics modeling. In recent years, artificial intelligence (AI) has demonstrated significant efficacy in the research and development (R&D) of electrode materials. Artificial neural networks (ANNs), convolutional neural networks (CNNs), deep neural networks (DNNs), generative adversarial neural networks (GANs), support vector machines (SVMs), and genetic algorithms (GAs) have been applied to design and optimize porous structures, compositions, materials, and surface properties for PEM fuel cells, demonstrating reliable and fast optimization and prediction capabilities. This article reviews the main physics and explores AI to advance porous electrode design for PEM fuel cells. Unlike traditional experimental and simulation-based approaches, AI provides superior computational efficiency, enabling faster and more cost-effective exploration of complex design parameters. In the end, future R&D directions for next-generation highly effective electrodes are discussed.