<p>This paper investigates the application of machine learning to predict the properties of pervious concrete. Traditional methods like lab tests and formulas have limitations. Machine learning offers a powerful alternative, analyzing mix design, non-destructive tests, and property measurements to predict properties like compressive strength and permeability. This approach can save time and resources by reducing lab testing and suggesting optimal mix designs faster. The present study analyzes the input variables, predicted properties, and specific machine learning algorithms employed and explores various machine learning algorithms and performance indicators for predicting pervious concrete properties. It details algorithm categories like tree-based methods, neural networks, and factors to consider when selecting a model. It also discusses performance indicators like R-squared and error distribution, emphasizing the importance of choosing appropriate metrics for a comprehensive evaluation. The findings suggest that Artificial Neural Networks (ANNs) and XGBoost models are generally the most effective for predicting key properties. The paper highlights the limitations of using AI for pervious concrete prediction, including data availability, model interpretability, material complexity, and environmental factors. Finally, it proposes solutions to these challenges, such as high-quality data collection, interpretable AI models, and collaborative research. This overview provides a valuable resource for understanding how machine learning can predict pervious concrete properties, ultimately optimizing mix design, reducing reliance on lab testing, and accelerating advancements in this field.</p>