Sub-symbolic AI focuses on enabling computers to learn directly from data by identifying complex patterns, contrasting with symbolic AI's reliance on explicit rules. This approach, inspired by the human brain, underpins advancements in areas like image and language processing through machine learning techniques, including supervised, unsupervised, and reinforcement learning. While powerful in pattern recognition and handling complex problems, sub-symbolic AI faces challenges in explainability, potential bias, and high computational demands, yet it drives numerous real-world applications and continues to evolve towards more transparent and efficient systems.