<![CDATA[General Background: The transition toward sixth-generation (6G) communication networks represents a shift from model-driven systems to intelligence-driven architectures centered on artificial intelligence (AI). Specific Background: AI techniques, including machine learning, deep learning, and reinforcement learning, are increasingly applied to network control and resource optimization tasks within highly dynamic and heterogeneous environments. Knowledge Gap: Existing research remains fragmented, with most studies addressing isolated network functions and lacking system-level integration, scalability considerations, and deployment feasibility. Aims: This study provides a system-level critical review of AI-driven control and optimization in 6G networks, evaluating capabilities, limitations, and architectural implications. Results: The analysis shows that while AI approaches improve adaptability and performance, they face challenges related to computational complexity, scalability, data constraints, interoperability, and limited explainability, alongside a clear gap between algorithmic advances and real-world implementation. Novelty: The study offers a structured synthesis, comparative evaluation of AI paradigms, and highlights the necessity of integrated and architecture-aware AI frameworks. Implications: The findings suggest that future 6G systems require hybrid AI models and unified frameworks to support scalable, reliable, and autonomous network operations, bridging theoretical innovation with deployment-oriented design. Highlights:• Identifies fragmentation of AI solutions across isolated network functions• Reveals trade-offs among learning paradigms in system-level deployment• Emphasizes need for integrated frameworks for scalable intelligent networks Keywords: Artificial Intelligence, 6G Communication Networks, Network Control, Resource Optimization, AI Native Architecture]]>
