AI for resource security: leveraging predictive analytics for sustainable management of water, energy, and food supplies in arid climates

Arid and semi-arid regions face intensifying pressure on limited water, energy, and food (WEF) resources as population growth and climate change amplify cross-sector dependencies. Artificial intelligence (AI), particularly predictive analytics, can support a shift from reactive resource management toward anticipatory forecasting, allocation, and maintenance. This study presents a systematic literature review of peer-reviewed research indexed in Scopus, Web of Science, and IEEE Xplore from January 2016 to March 2024. After deduplication, 842 unique records were screened, 157 full texts were assessed, and 29 empirical studies focused on arid or semi-arid contexts were included. The review maps AI methods, applications, datasets, regional contexts, performance measures, and implementation constraints across the WEF nexus. Long Short-Term Memory (LSTM) and related deep-learning models dominate time-series forecasting, while optimisation and digital-twin approaches support irrigation, water allocation, desalination, network operation, and energy-management decisions. The synthesis identifies a scalability paradox: regions with the greatest potential benefit frequently have the lowest adoption capacity because of sparse data, weak digital infrastructure, institutional fragmentation, cost, skills gaps, and limited trust in opaque models. A multi-level framework is proposed that links open and interoperable data, low-bandwidth and explainable AI, local co-design, capacity development, and cross-sector governance. Technical performance is necessary but insufficient; equitable resource security depends on aligning AI deployment with context-specific data, institutional readiness, and ethical safeguards.