The increasing complexity and scale of IoT networks demand intelligent, adaptive network control mechanisms to ensure efficiency, scalability, and robustness. Traditional rule-based approaches are insufficient to handle the dynamic nature of IoT systems, necessitating AI-driven solutions. This chapter explores how Deep Reinforcement Learning (DRL), Graph Neural Networks (GNNs), and Large Language Models (LLMs) revolutionize IoT network management by enabling real-time adaptation, decentralized decision-making, and human-like interaction with network systems. The chapter first introduces the fundamental challenges of IoT network control, including dynamic topologies, fluctuating traffic patterns, and interference management. It then presents DRL-based adaptive monitoring, which optimizes the significant sampling problem, ensuring efficient data collection while minimizing energy consumption and network overhead. Next, GNN-based decentralized control is explored, demonstrating how AI enhances multi-device collaboration, interference mitigation, and dynamic resource allocation. Finally, LLM-driven network orchestration is introduced, illustrating how natural language interfaces can bridge human input with AI-powered IoT control, making network management more accessible. Through extensive case studies and numerical analyses, this chapter showcases the impact of AI-enhanced control in improving network efficiency, reducing communication delays, and enhancing fault tolerance. The discussion extends to real-world IoT applications, including smart cities, industrial automation, and autonomous systems.