Recent breakthroughs in machine learning have revolutionized numerous fields, with healthcare standing out as an especially promising beneficiary. Clinical electroencephalography (EEG) classification is one task in which the development of machine learning techniques would be particularly beneficial, to support the small number of highly trained interpreters on which this field depends. Delving into the cutting-edge developments in machine learning applications for EEG classification, this text challenges the conventional performance ceilings previously perceived as unattainable. It begins with an essential overview of EEG’s role in clinical diagnostics, followed by an analysis of the challenges inherent in EEG data interpretation and the innovative machine learning approaches designed to address them. A discussion on the critical issue of data scarcity in EEG analysis introduces a multi-faceted strategy to enhance data utility and model performance, examination of morphological techniques like window length extension, and second-stage model arbitration. The exploration extends to the potential integration of diverse EEG model architectures, particularly those inspired by audio classification advancements, providing new perspectives on EEG signal analysis. Looking forward, the text emphasizes the need to reevaluate performance benchmarks and explore multimodal learning approaches for a holistic understanding, aiming to push the boundaries of what’s achievable in clinical EEG interpretation with AI technologies.