Uncovering battery electrochemical mechanisms by artificial intelligence

Batteries have been driving the sustainable energy transition by empowering critical applications such as consumer electronics, electric vehicles and grid energy storage systems. Key challenges in battery research and development require a fundamental understanding of the dynamic evolution of electrochemical interfaces, cross-dimensional and cross-scale relationships, and intertwined interaction electrochemical processes. Advanced characterization and theoretical computation-based methods generate considerably discrete, heterogeneous and condition-sensitive but huge data streams. Such complexity leads to difficulties in human expert-oriented interpretations. Artificial intelligence (AI) offers new promise for handling this gigantic amount of data by enabling efficient curation, preprocessing, model construction, deployment, optimization and, most importantly, interpretation. While AI integration into battery research has been well documented, this Review pays special attention to its potential to uncover three critical yet outstanding chemical mechanistic aspects. First, AI reveals temporal evolution mechanisms by denoising and statistically analyzing large, uneven-quality time-resolved data. Second, it discovers latent relationships across data with multiple dimensions and scales, which are difficult to infer from established theories alone. Third, it decouples complex interaction networks by identifying dominating factors and their relative contributions. We highlight the importance of standardized data collection, open-source data deposition, domain expert knowledge integration, application of advanced AI models, and experiment optimization to scalable and electrochemistry-informed AI applications. While emerging tools like large language models and autonomous agents hold promise, their impact will rely on thoughtful human-AI collaboration that preserves safety, ethics and mechanistic insight.