Genome-scale metabolic modeling and simulation have been widely employed in biological studies and biotechnological applications due to their powerful capabilities of estimating metabolic fluxes at the systems level. In recent years, machine learning (ML) has been beginning to be applied to the reconstruction and analysis of genome-scale metabolic models (GEMs) to improve their quality. Also, ML has been used to diversify the utilization of information derived from genome-scale metabolic modeling and simulation. Recent studies have shown that machine learning can improve predictive performance and data coverage of GEMs. Also, genome-scale metabolic modeling and simulation provide interpretability of ML applications. Although many biological data still need to be made suitable for ML applications, it is expected that ML will be increasingly applied to GEMs to further improve the practical use and find new applications of GEMs.