Sustainability paradoxes in AI-enabled decentralized energy systems: Opportunities, risks, and systemic traps

Decentralized energy systems (DESs) can localize renewable energy conversion, improve resilience, and strengthen community participation in the energy transition. AI is increasingly used in DESs for forecasting, optimization, multi-agent coordination, and fault detection, enhancing efficiency and operational flexibility. However, this review shows that AI also introduces significant sustainability trade-offs. Across the literature, AI-enabled DESs face tensions linked to the efficiency–sustainability, optimization–resilience, participation–exclusion, and autonomy–agency paradoxes. These tensions are reinforced by data limitations, weak standardization, high computational and hardware demands, and insufficient attention to life-cycle impacts. This structured critical review examines the opportunities, challenges, and sustainability paradoxes associated with AI in DESs and links them to real-world implementation barriers. Based on this analysis, it identifies key mitigation pathways, including lifecycle-based assessment, lightweight and resource-efficient AI, circular hardware design, multi-objective optimization with resilience and equity criteria, open interoperable platforms, and participatory governance. The review argues that AI is not inherently sustainable; its contribution depends on how it is designed, deployed, and governed. By integrating technical, environmental, economic, and social perspectives, the paper offers a structured framework for developing AI-enabled DESs that are more resilient, equitable, and aligned with long-term sustainability goals.