Global Air Traffic Management (ATM) faces unprecedented challenges as flight density increases in congested airspace. Traditional systems rely on human monitoring supported by radar, an approach that may encounter difficulties when dealing with sudden weather changes or the accumulation of flights. This is where computer science plays a vital role through the application of recurrent neural networks, particularly the Long Short-Term Memory (LSTM) architecture, which is specifically designed to process sequential and time-dependent data. How the Computational System Works: LSTM algorithms are trained on large amounts of historical and real-time data, including: - Previous flight trajectories and delay rates. - Changing weather forecasts and wind movements. - Current flight plans and takeoff and landing schedules. The neural network processes these inputs to predict potential air traffic congestion before it occurs, with forecasts extending up to two hours, depending on the system and the available data. The algorithm does not simply identify potential problems; it can also suggest alternative routes and optimized aircraft speeds. These recommendations are then presented to air traffic controllers, who make the appropriate decisions. Results and Impact: These systems can help reduce airborne waiting times and improve the flow of air traffic. However, the level of improvement depends on operating conditions, data accuracy, and the effectiveness of the model used. Optimizing flight routes and reducing delays can also lower fuel consumption and carbon emissions, supporting the development of smarter, more efficient, and more sustainable air traffic management systems.