Accelerating sustainable energy storage and conversion with AI and ML: A perspective on emerging directions

Achieving global decarbonization goals by 2050 demands rapid progress in renewable energy and energy storage and conversion technologies. Artificial Intelligence (AI) and Machine Learning (ML) are increasingly used in Materials Science, enabling faster and more efficient discovery, characterization, and optimization of energy storage and conversion materials. Drawing on insights from the 2024 MRS Energy & Sustainability Panel, this perspective explores how AI/ML is reshaping the energy materials landscape, from predictive modeling and microstructural analysis to autonomous laboratories and closed-loop experimentation. It highlights six key areas of progress and challenge, including data infrastructure, sustainability of AI itself, and the need for interdisciplinary collaboration, and it argues that the rate-limiting step is now the evidence rather than the algorithm: claims of AI-accelerated discovery should be assessed on data readiness, model uncertainty, experimental validation, autonomy level, and life-cycle benefit. By integrating literature examples and expert viewpoints, the article presents an overview of how AI and ML can accelerate innovation in batteries, catalysts, and emerging material systems, while emphasizing the importance of open data, reproducibility, and responsible deployment to ensure long-term impact and sustainability. Artificial intelligence (AI) is accelerating the discovery of advanced materials for batteries, fuel cells, and other clean-energy technologies. However, the reliability and practical value of AI-driven materials discovery depend critically on the quality of the underlying data, the robustness of model predictions, and experimental validation. We propose a framework for evaluating AI-driven materials discoveries based on five key considerations: data quality, predictive uncertainty, experimental synthesis and validation, degree of automation, and the energy and environmental costs associated with AI-driven approaches. Together, these criteria provide a basis for assessing not only whether an AI system can identify promising materials, but also whether the resulting discoveries are reproducible, experimentally validated, and environmentally meaningful.