Research on Task Scheduling in Cloud Computing focuses on developing efficient methods to optimize resource allocation, reduce execution time, and improve energy efficiency in dynamic cloud environments. Recent studies highlight AI-driven scheduling models that leverage machine learning, heuristic, and hybrid techniques to handle resource heterogeneity and real-time adaptability. Hybrid optimization approaches, such as combining Particle Swarm Optimization with Grey Wolf Optimization, or integrating Transformer models with Cuckoo Search, enhance convergence speed and solution quality. Deep Reinforcement Learning methods provide adaptive job scheduling and resource management, while predictive graph networks support dynamic load balancing in heterogeneous cloud systems. Collectively, these advancements enable more intelligent, scalable, and efficient task scheduling frameworks for modern cloud computing environments.