The work aims to sort out the current research status of emotion-driven AI art from three aspects: data, methods, and applications, and summarize the development history, limitations, and future research trends in this field. First, based on the literature, research results, and application products, the changes in emotion annotation methods and their impacts on affective art databases were reviewed. Then, the implementation principles of technologies such as emotion recognition, controllable generation, and cross-media search were explained, and a variety of emotion-driven AI artistic methods were discussed. Finally, three interaction mechanisms between emotion and AI art were summarized, and the impact of related products on creativity, healthcare, education, and business marketing was analyzed. Affective computing methods enhance the infectiousness of AI artworks and support human-AI co-creation. With the further integration of art and technology, the art field will usher in new modes of production, experience, and consumption.