Conventional interpretation of airborne electromagnetic data has been conducted by solving the inverse problem. However, with recent advances in machine learning (ML) techniques, a one-dimensional (1D) deep neural network inversion that predicts a 1D resistivity model using multi-frequency vertical magnetic fields and sensor height information at one location has been applied. Nevertheless, bacause the final interpretation of this 1D approach relies on connecting 1D resistivity models, 1D ML interpretation has low accuracy for the estimation of an isolated anomaly, as in conventional 1D inversion. Thus, we propose a two-dimensional (2D) interpretation technique that can overcome the limitations of 1D interpretation, and consider spatial continuity by using a recurrent neural network (RNN). We generated various 2D resistivity models, calculated the ratio of primary and induced secondary magnetic fields of vertical direction in ppm scale using vertical magnetic dipole source, and then trained the RNN using the resistivity models and the corresponding electromagnetic (EM) responses. To verify the validity of 2D RNN inversion, we applied the trained RNN to synthetic and field data. Through application of the field data, we demonstrated that the design of the training dataset is crucial to improve prediction performance in a 2D RNN inversion. In addition, we investigated changes in the RNN inversion results of field data dependent on the data preprocessing. We demonstrated that using two types of data, logarithmic transformed data and linear scale data, which having different patterns of input information can enhance the prediction performance of the EM inversion results.