The increased use of social media platforms, which provide user-generated data in real-time, has significantly impacted the disaster response and recovery field. The application of artificial intelligence (AI) techniques to parse multimodal datasets from social media has emerged as a practical and feasible strategy to enhance disaster management efficacy in response to this paradigm shift. This research provides a thorough overview of the most recent advancements in AI-driven social media data analysis methods to aid in disaster response and recovery operations. The study explains the wide range of available AI-driven solutions, covering state-of-the-art techniques such as computer vision, deep learning, machine learning, and natural language processing (NLP). This study presents a review based on 44 recent papers, 32% of which are survey papers, 64% on an implementation approach and 4% case study. Survey papers selected for the comprehensive review provide valuable insights covering 1715+ papers. The findings offer valuable input related to the dataset used in recent research, the proposed methodology, and future directions. The research makes a significant contribution to synthesizing the findings for future work in the proposed area.