Microstrip Patch Antennas (MPAs) are increasingly ubiquitous elements of modern wireless communication systems thanks to their small size, compactness, and straightforward production processes. However, conventional antenna design techniques face serious drawbacks such as low bandwidth, insufficient gain, and large computation costs associated with EM simulations. Advances in the field of AI and Machine Learning have led to the emergence of data-driven antenna design processes that greatly improve performance and efficiency of antenna development. This study offers an extensive review of state-of-the-art research into AI-driven optimization techniques used for designing microstrip patch antennas. Various Machine Learning models, including Artificial Neural Networks (ANN), Convolutional Neural Networks (CNN), Genetic Algorithm (GA), Particle Swarm Optimization (PSO), and hybrid optimization techniques are analyzed. Such approaches allow to obtain significant improvements in antenna performance and efficiency, providing a better bandwidth, higher gain, lower return loss, etc., while significantly decreasing computational efforts [2], [5], [9]. In addition, emerging trends associated with inverse design and ML-driven hybrid optimization are considered [6], [7].