Researchers at the University of New Hampshire have used artificial intelligence to dramatically speed up the discovery of new functional magnetic materials, creating a comprehensive, searchable database of 67,573 magnetic compounds—25 of which were previously unknown and remain magnetic at high temperatures. This breakthrough could help reduce reliance on rare earth elements, which are critical but costly and often imported materials used in permanent magnets. Suman Itani, a doctoral student in physics and the lead author of the study, emphasized the broader impact: “By accelerating the discovery of sustainable magnetic materials, we can reduce dependence on rare earth elements, lower the cost of electric vehicles and renewable-energy systems, and strengthen the U.S. manufacturing base.” The newly developed database, called the Northeast Materials Database, catalogs magnetic materials essential to modern technology—ranging from smartphones and medical devices to power generators and electric vehicles. Despite the widespread use of magnets, no new permanent magnet has been discovered in decades, even though countless magnetic compounds are theoretically possible. The team’s approach leveraged AI to analyze scientific literature, extracting key experimental data such as chemical composition, crystal structure, and magnetic properties. This information was then used to train machine learning models capable of predicting whether a material is magnetic and how stable its magnetism is at elevated temperatures. “Testing every possible combination of elements—potentially millions—through traditional lab experiments is simply too slow and expensive,” said Jiadong Zang, a physics professor and co-author. “We are tackling one of the most difficult challenges in materials science—discovering sustainable alternatives to permanent magnets—and we are optimistic that our experimental database and growing AI technologies will make this goal achievable.” The research, published in Nature Communications, also highlights the potential of the underlying large language model to extend beyond materials science. Co-author Yibo Zhang, a postdoctoral researcher in both physics and chemistry, noted that the same AI framework could be adapted to modernize academic library collections by converting legacy images and documents into accessible digital formats, offering broad applications in education and research infrastructure.