The adoption of electric vehicles (EVs) is increasing worldwide as it may help reduce reliance on fossil fuels and greenhouse gas emissions. However, the large-scale use of charging stations for electric vehicles poses some challenges to the power grid and public infrastructure. To overcome the problem of extended charging time, the simple solution of increasing the charging station and increasing the charging capacity does not work due to the load and space limitation of the power grid. Therefore, researchers focused on developing intelligent planning algorithms to manage the demand for public charging based on predicting the charging time of electric vehicles. As a result, this paper proposes a deep learning approach for predicting the duration of charging sessions. These approaches are validated using a real-world dataset of charging processes collected at public charging stations in Morocco. Numerical results show that the gated recurrent units (GRU) regression method slightly outperforms the other methods in predicting the charging sessions duration. Accurate prediction of electric vehicles charging duration has many potential applications for utilities and charging operators, including grid reliability, scheduling, and smart grid integration. In the case of Morocco, the massive deployment of EVs can cause a variety of problems to the electrical system due to the considerable charging power and stochastic charging behaviors of electric vehicle drivers. Thanks to this study's results, we can assess the expected impact of additional EVs on the grid, considering specific characteristics of the Moroccan power system.