Revolutionizing supramolecular materials design with artificial intelligence

The design and development of supramolecular materials are hindered by complex non-covalent interactions and a lack of comprehensive rational design theories. Traditional "trial-and-error" methods are inefficient and labor-intensive, slowing progress in creating materials with precise tunability, robust stability, multifunctionality, and dynamic behavior. This perspective highlights major difficulties in supramolecular materials research and the transformative potential of artificial intelligence (AI) and machine learning (ML) in revolutionizing the field. Key challenges in applying AI include limited data availability, data quality issues, and the path-dependent nature of assembly processes. To overcome data scarcity, we discuss strategies such as transfer learning, data augmentation, and federated learning to enhance model performance with small datasets. We propose developing Intelligent Data Manufacturing Platforms—advanced laboratory automation systems designed to generate large volumes of high-quality data. By integrating AI algorithms with robotics in a closed-loop experimental system, these platforms enable high-throughput experimentation, autonomous decision-making, and iterative refinement of AI models through continuous data acquisition. This accelerates the design-build-test-learn cycle, fostering innovation and facilitating the development of next-generation supramolecular materials. By establishing standardized data repositories and encouraging global collaboration, this framework propels the field toward a data-intensive paradigm.