This paper reviews how artificial intelligence and electroencephalography (EEG) are being combined to improve the prediction and diagnosis of Major Depressive Disorder (MDD). The authors carried out a systematic review of studies from 2015 to 2024, looking at how EEG signals are processed and analyzed with machine learning and deep learning models. They found a wide range of approaches to preprocessing EEG signals, removing artifacts, and extracting features, with convolutional neural networks and hybrid deep learning models often outperforming traditional methods like support vector machines. While some models reported very high accuracy (sometimes above 90%), results were inconsistent due to differences in study design, datasets, and evaluation methods. This inconsistency makes it difficult to compare findings across studies and slows down the translation of these techniques into clinical practice. The authors conclude that more standardized protocols, larger and more diverse datasets, and explainable AI models will be necessary to make EEG-based AI tools reliable and clinically useful.