An international team of physicists has developed a new method using machine learning to identify chemical combinations that could lead to superconductors—materials capable of conducting electricity without resistance. This approach has already led to the discovery of two new superconducting materials: YRu3B2 and LuRu3B2. These materials have unique electronic structures with flat bands, a feature often linked to a special geometric pattern known as the kagome lattice. This pattern resembles the interwoven design of traditional Japanese baskets and is of particular interest in physics because it can lead to unusual electronic behaviors. The research was conducted by the international collaboration known as SuperC, which was launched in 2023 and is led by Aalto University in Finland. The project combines principles from quantum physics, the study of electronic structures, and artificial intelligence to speed up the search for new materials. Instead of testing every possible chemical combination through expensive and time-consuming simulations, the team used a machine learning algorithm to first eliminate less promising options. This allowed them to focus their more precise quantum calculations on the most promising candidates. Once the digital screening was complete, the selected materials were synthesized in the laboratory by researchers from Rice University in the United States. Experimental tests on the resulting compounds confirmed their superconducting properties. Superconductors are remarkable because they can carry electric current without energy loss, but most require extremely low temperatures to function—often near absolute zero. This makes them impractical for widespread use without costly cooling systems. This breakthrough demonstrates that machine learning can play a transformative role in the discovery of superconductors. Historically, the search for new superconducting materials has relied heavily on trial and error or unpredictable experimental results. While over 7,000 superconducting materials have been identified, very few were predicted accurately before being made in the lab. By using AI to filter and prioritize the most promising candidates, researchers can now explore vast numbers of material combinations that were previously too computationally expensive to analyze. In the long term, this method could greatly increase the chances of discovering more practical superconductors, potentially even ones that work at room temperature, which would revolutionize technology and energy efficiency.