In this paper, we have created an interparticle interaction potential for liquid alkali metals using ab-initio simulations and machine learning. The resulting family of potentials was tested by molecular dynamics simulation, in which the liquid metals were heated to the melting point and the resulting system was then averaged. The simulation results show that near the melting point, the radial distribution function and the static structure factor are reproduced. This results indicates that the principle of corresponding states is fulfilled in liquid alkali metals.