As the global transition to renewable energy intensifies, the search for high-performance batteries and efficient electrocatalysts has become a critical race against time. Traditionally, discovering these materials required years of "trial-and-error" laboratory experiments. Now, a comprehensive review published in the journal ENGINEERING Energy by researchers from Tongji University reveals how Artificial Intelligence (AI) is fundamentally shifting this paradigm.
The study, led by Professor Menghao Yang's team at the Institute of New Energy for Vehicles, Tongji University, provides a systematic roadmap of the AI evolution in energy materials—tracing a complete technical pathway from classical Machine Learning (ML) to Representation methods, then to Discriminative tasks, followed by Generative tasks and Domain-integrated AI systems, and finally to Large Models.
Beyond Trial and Error: The Rise of "Inverse Design" One of the most transformative shifts highlighted in the research is the move toward "Inverse Design." Unlike traditional methods that test existing materials to find their properties, AI-driven generative models allow scientists to start with a desired performance goal (such as high energy density or specific catalytic activity) and work backward to predict the exact chemical structure required.
"The integration of AI into energy materials research is no longer just a trend; it is a necessity for efficiency," says Professor Yang. "By utilizing generative AI and Large Language Models, we can navigate the vast chemical space of potential materials at speeds that were previously unimaginable."
Batteries, Catalysts, and the Power of Large Models The review delves into two primary domains where AI is making the most significant impact:
- Secondary Batteries: AI algorithms are now used to predict battery lifespan, optimize electrolyte compositions, and enhance the safety of Li-ion and next-generation battery systems.
- Electrocatalysis: For critical reactions like the Hydrogen Evolution Reaction (HER) or Oxygen Reduction Reaction (ORR), AI helps identify optimal surface structures of catalysts, facilitating the production of green hydrogen and the reduction of CO2 emissions.
The researchers emphasize that the emergence of "Large Models" (including LLMs) is particularly exciting. These models can process vast amounts of unstructured scientific literature, extract hidden correlations, and even suggest new experimental synthesis routes, acting as an "intelligent co-pilot" for material scientists.
Future Horizons While the potential is vast, the team notes that challenges remain, particularly regarding the quality of experimental data and the "black box" nature of some AI models. The paper outlines a future where "Self-Driving Laboratories"—where AI designs, performs, and analyzes experiments autonomously—could become the standard for energy research.
JOURNAL: ENGINEERING Energy
DOI: https://doi.org/10.1007/s11708-026-1053-5
Article Link: https://link.springer.com/article/10.1007/s11708-026-1053-5
Cite this article: Jiang, M., Zhou, J., An, Y., Lin, Z., & Yang, M. (2026). Artificial intelligence for energy materials research: From classical machine learning to large models. ENGINEERING Energy, 20(1), 10535. https://doi.org/10.1007/s11708-026-1053-5
