Abstract Artificial intelligence (AI) is playing an ever-growing role in materials research; however, its full potential for promoting sustainable development has yet to be realized1. Current materials development is still largely guided by the traditional linear framework of design, production, use, and disposal2. Although AI has improved how fast and accurately we can predict material properties and screen materials, these improvements often focus more on performance and cost than on life-cycle assessment (LCA) and the ability to recover resources, which can negatively impact the environment3. To address these challenges, Han et al. from the National Institute of Clean-and-Low-Carbon Energy have proposed a more integrated framework, structured around five strategic pillars: revisiting research and development (R enabling AI-driven circular design; establishing closed-loop digital twin systems; integrating LCA methodologies; and building system-level R&D platforms4. The objective is to reposition AI as a core driver of the entire material life cycle rather than merely a discovery tool. This paradigm shift allows for the simultaneous optimization of efficiency, resource recovery, and environmental performance, thereby accelerating the development of sustainable materials with high efficiency and low carbon emissions.
Zhijie Fu (2026) studied this question.