The rapid advancement of Vehicle-to-Everything (V2X) communication is transforming Intelligent Transportation Systems (ITS), with 6G networks expected to provide ultra-reliable, low-latency, and high-capacity connectivity for Connected and Autonomous Vehicles (CAVs). Artificial Intelligence (AI) and Machine Learning (ML) have emerged as key enablers in optimizing V2X communication by enhancing network management, predictive analytics, security, and cooperative driving due to their outstanding performance across various domains, such as natural language processing and computer vision. This survey comprehensively reviews recent advances in AI and ML models applied to 6G-V2X communication. It focuses on state-of-the-art techniques, including Deep Learning (DL), Reinforcement Learning (RL), Generative Learning (GL), and Federated Learning (FL), with particular emphasis on developments from the past two years. Notably, AI, especially GL, has shown remarkable progress and emerging potential in enhancing the performance, adaptability, and intelligence of 6G-V2X systems. Despite these advances, a systematic summary of recent research efforts in this area remains lacking, which this survey aims to address. We analyze their roles in 6G-V2X applications, such as intelligent resource allocation, beamforming, intelligent traffic management, and security management. Furthermore, we explore the technical challenges, including computational complexity, data privacy, and real-time decision-making constraints, while identifying future research directions for AI-driven 6G-V2X development. This study aims to provide valuable insights for researchers, engineers, and policymakers working towards realizing intelligent, AI-powered V2X ecosystems in 6G communication.