Imagine effortlessly exploring complex datasets, uncovering hidden patterns with just a simple question. That's the promise of Large Language Model (LLM)-assisted visual analytics, a field poised to revolutionize how we interact with and understand data. Traditionally, visual analytics has required specialized expertise, creating a barrier for non-technical users. Now, LLMs are breaking down these barriers, transforming data exploration into a natural, intuitive conversation. This emerging field integrates the power of LLMs into every facet of visual analytics—from data management and visualization generation to user guidance and multimodal interaction. Current research highlights how LLMs are streamlining data preprocessing, making it possible to analyze raw, unstructured data with ease. They are also enabling the generation of visualizations directly from natural language queries, eliminating the need for complex coding or specifications. The ability of LLMs to leverage vast domain knowledge opens doors to insightful analyses across diverse fields. Furthermore, the integration of multimodal interactions—combining language with touch, gestures, or gaze—promises a richer, more engaging analytical experience. But the journey isn't without its challenges. Ensuring data accuracy, maintaining user trust through explainable AI, and effectively evaluating these complex systems are critical hurdles. Moreover, limitations in LLMs' reasoning abilities, especially with mathematical or numerical data, require further research and development. Despite these challenges, the potential of LLM-assisted visual analytics is undeniable. By addressing these obstacles head-on, we can unlock the full potential of this transformative technology, empowering everyone to explore data, uncover insights, and make better data-driven decisions.