<p>This review article explores the emerging field of Chaotic Neural Networks (cNNs), which integrates chaos theory with neural network paradigms to provide a novel framework for artificial intelligence and computational neuroscience. cNNs are classified into various categories based on where chaos is incorporated into the architecture, and their capabilities, such as the ability to capture complex, non-linear dynamics and perform tasks like pattern recognition, optimisation, and time-series prediction, are reviewed. Applications of each class are also discussed. The challenges associated with integrating chaos into neural networks, such as the sensitivity of chaotic systems to initial conditions and the necessity of effective control mechanisms to maintain stability, are addressed. A comparative analysis of traditional machine learning algorithms and cNN is provided, which highlights their advantages in managing complex tasks while also recognising their limitations, such as their susceptibility to vanishing gradients and computational complexity. Future research directions are proposed, emphasising the refinement of chaotic dynamics, the improvement of interpretability, and the establishment of standardised benchmarks for the evaluation of cNNs.</p>