Over the years, molecular dynamics (MD) simulations have been employed in the study of carbohydrates, with force fields such as CHARMM, AMBER/GLYCAM, and GROMOS. Although these force fields have achieved considerable success and played a pivotal role in our understanding of carbohydrate chemistry, growing interest has emerged in incorporating polarization effects to enhance the accuracy of simulations. In this perspective, we contemplate the advances that have been made in nonpolarizable and polarizable force fields to extract the key factors controlling accuracy in MD of carbohydrates. We find that the extreme hydrophilicity and conformational flexibility of carbohydrates pose challenges for most force fields. Overall, a force field suited for carbohydrates needs to include a water model developed consistently with the solute parameter sets, a soft van der Waals repulsion term at short distances, and polarization (whether implicit or explicit). We find that AMOEBA improves the prediction of hydration shell structure and dynamics, hydrogen bonding, and kinetics of diffusion, although it remains largely untested for conformational flexibility and glycosidic linkages. Nevertheless, AMOEBA’s recent success in modeling monosaccharides without revisions of the potential energy functions or water model presents a promising avenue for future research. Such advances will provide deeper insights into the structure, dynamics, and interactions of these biologically and industrially relevant macromolecules.