Earthquakes are among the most destructive natural disasters that constantly threaten human life. Therefore, taking preventive measures is essential. Although Artificial Intelligence (AI) can potentially improve Earthquake Disaster Risk Management (EDRM), its current applications, challenges, and research gaps remain insufficiently explored. This study investigates the application of AI in earthquake-related disaster risk management (AI-EDRM). It systematically analyses the distribution of DRM phases and sub-phases, AI types, subfields, and problem categories. Moreover, it evaluates commonly used AI applications and algorithms, examines data types and methodological approaches in earthquake studies, and identifies significant challenges and research gaps. To this end, a systematic review of 55 articles indexed in Scopus and Web of Science was conducted. Python was used for data processing and visualisation. Findings reveal that AI-EDRM research has become increasingly diversified. However, while most studies focus on post-disaster response, the recovery phase remains underexplored. The most commonly studied AI-EDRM sub-processes include damage assessment, forecasting and prediction, whereas planning and risk awareness are neglected. The main challenge is the lack of high-quality and well-integrated data. Key research gaps include limited integration of heterogeneous data sources and underutilisation of advanced AI models. Future directions emphasise the development of generalised datasets, multimodal data fusion, advanced AI architectures, and hybrid approaches. Moreover, there is a need for decision support systems aligned with local governance strategies, incorporating explainable AI (XAI) and fuzzy inference systems (FIS) to enhance transparency and manage uncertainty.