<p>Slopes under conditions such as seismic loading or rainfall are highly susceptible to instability phenomena including landslides, rockfalls, debris flows and so on. Traditional methods such as the limit equilibrium method and numerical simulation offer simplicity and intuitive application but often struggle to accurately identify the most critical slip surface and compute the accurate and effective factor of safety (<i>FOS</i>) when dealing with heterogeneous soil layers, dynamic loads, and complex geological conditions. These limitations may lead to discrepancies between predicted results and actual engineering performance. To address these challenges, researchers have increasingly introduced intelligent algorithms such as machine learning and neural networks, aiming to improve the accuracy and efficiency of slope stability analysis through data-driven approaches and intelligent optimization techniques. This review presents: (1) A comprehensive review of advances in slope stability analysis over the past decade is presented, covering both conventional and intelligent methods. Key limitations of intelligent approaches are identified, including sensitivity to parameter selection, limited model generalization, data sparsity, challenges in multi-source data integration, and a lack of extensive field validation and standardized frameworks. (2) a critical analysis of existing challenges in intelligent methods, including parameter sensitivity, model generalization, data sparsity, and multi-source data fusion, along with potential solutions; (3) prospects for future development, including multi-physics coupling modeling, adaptive learning systems, explainable artificial intelligence techniques, and standardized data platform construction, to provide theoretical support and practical insights for solving complex slope stability problems in engineering practice.</p>