Mushroom classification is of significant importance in agriculture, ecology, and food safety; however, accurate classification is non-trivial, as mushrooms exhibit considerable morphological variation, high inter-class similarity, and a shortage of large-scale annotated data. Recent developments in Artificial Intelligence (AI), specifically in Machine Learning (ML) and Deep Learning (DL), have brought powerful remedies to these concerns. This study provides a comprehensive overview of ML and DL techniques in the context of mushroom classification and edibility prediction. We compare trained classical ML techniques and models, such as decision trees, random forests, support vector machines, and ensemble models, to modern DL architectures (e.g., CNNs, transfer learning models, and lightweight network design optimal for mobile) in the context of these objectives. The survey also reviews hybrid approaches, object detection models, and data balancing methods, and presents how they affect classification performance. Cross-dataset comparison suggests that, under controlled conditions, ML and DL can achieve nearly perfect accuracy, while challenges such as generalization, small datasets, and class imbalance are observed. We conclude the paper by discussing critical challenges, such as large-scale curated datasets and resistance to environmental variations, and then present the future directions of this burgeoning field, including multimodal fusion, real-time mobile applications, and domain adaptation techniques. This survey serves as a quick reference for researchers and practitioners who aim to develop intelligent mushroom recognition systems.