Sixth-generation (6G) wireless networks are entering the standardisation horizon for the early 2030s, with research roadmaps already published by 3GPP, ITU-R, and major industrial alliances. Unlike previous generations, 6G is being designed as an AI-native system in which machine learning is integrated into the air interface, the radio access network, and the core. This paper surveys the architectural vision of 6G, key enabling technologies sub-terahertz spectrum, reconfigurable intelligent surfaces (RIS), cell-free massive MIMO, AI-RAN, network digital twins, and semantic communications and the role of artificial intelligence at each layer. We discuss energy efficiency, spectrum sharing, integrated sensing and communication, and the security implications of pervasive learning components. Open research problems related to standardisation, explainability, and sustainable deployment are highlighted.