Abstract: Recent advances in artificial intelligence have revolutionized the structure prediction and functional design of proteins, providing novel methods and tools for the de novo design of artificial enzymes. Serine hydrolases have emerged as a representative model in the task of artificial enzyme design due to their well-defined catalytic mechanism. This review summarizes key deep learning algorithms recently developed for artificial enzyme engineering and highlights their typical applications, with a focus on the latest successes in designing artificial serine hydrolases. These breakthroughs pave the way for the development of artificial enzymes with enhanced catalytic efficiency and stability, marking a new stage in enzyme design that could transform biomanufacturing and related industries.