Laser polishing (LP) is a novel non-additive and non-subtractive manufacturing technology that enables the addition of augmented surface functionalities such as enhanced surface quality, visual appearance, wettability, and friction. The main goal of the current study is represented by the development of superior control systems capable to rely on artificial intelligence methods in order to predict the resultant surface quality. For this purpose, the effect of recurrent neural networks (RNN)’s ability in modelling the time/space domain dynamics of the LP was evaluated by comparing its performance with that predicted by feed-forward neural networks (FFNN). It was observed that RNN’s were able to achieve much higher prediction accuracy with a correlation coefficient of 0.78 and 0.99 for areal waviness and roughness, respectfully.