Gesture recognition has emerged as a pivotal component in the development of intuitive and immersive Augmented Reality (AR) and Mixed Reality (MR) systems. This study explores the application of Artificial Intelligence (AI) techniques to enhance gesture recognition accuracy, efficiency, and usability in real-time AR/MR environments. We present an experimental study leveraging state-of-the-art deep learning architectures, evaluating their performance in recognizing gestures across diverse datasets and AR/MR platforms. The experiments focus on comparing various AI models under different environmental conditions, such as lighting variations, occlusions, and gesture complexities. Results indicate significant improvements in recognition rates and responsiveness, offering a robust foundation for future interactive systems. Challenges and opportunities for future research are also discussed.