We developed a polarizable force field to investigate mixtures of choline chloride. Dynamic properties are in good agreement with experimental values. However, the model lack of transferability between different systems since a system specific damping function for interactions of the anion must be introduced. Furthermore, this parameter must be fitted to expensive first principle molecular dynamics simulations. Therefore, we have investigated the capabilities of machine learning inter-atomic potentials. This enables molecular dynamics simulations of systems consisting of several thousand of atoms on the ns-time scales with quantum chemistry accuracy but at a fraction of the computational cost of ab initio molecular dynamics simulations.