In this work, the application of Convolutional Neural Networks (CNN) for improving Adaptive Modulation and Coding (AMC) in 5G communication systems is explored. AMC optimizes data transmission under uncertain channel conditions by dynamically selecting modulation schemes and correcting errors. Traditional AMC relies on manual feature extraction, which is time-consuming and may overlook critical patterns. To address this, a structured CNN pipeline with convolutional layers, activation functions, and pooling techniques is proposed to spatially extract features from in-phase and quadrature-phase signal components. The model demonstrates robust classification across varying Signal-to-Noise Ratio (SNR) levels using modulation schemes like BPSK, QPSK, 16QAM, 64QAM, and 256QAM. Experimental results show that the proposed CNN model achieves an accuracy of 95%, indicating strong generalization to unseen data. This approach underscores the CNN’s potential to automatically learn hierarchical features from raw input, enhancing the reliability and efficiency of 5G wireless communication systems.