The U.S. National Science Foundation (NSF), in partnership with Intel, will invest $20 million over five years to establish the Artificial Intelligence Materials Institute (NSF AI-MI) at Cornell, as part of the National Artificial Intelligence Research Institutes. The NSF announced the investment on July 29.
Directed by Eun-Ah Kim, principal investigator (PI) and the Hans A. Bethe Professor of physics in the College of Arts and Sciences (A&S), NSF AI-MI will accelerate and transform the discovery of new materials to be used in sustainable energy, advanced electronics, environmental stewardship and quantum technologies by integrating human scientific expertise with AI methods.
Researchers from A&S, the Cornell Ann S. Bowers College of Computing and Information Science and Cornell Engineering make up the majority of the new institute’s leadership team, joined by researchers from Princeton University; the City College of the City University of New York (CUNY); the Advanced Science Research Center at CUNY; and Boston University.
“Materials move and change the world,” Kim said. “However, one thing we haven’t been able to do yet is intentionally design and synthesize material.”
While past materials research relied on serendipitous discovery, there is now a movement toward targeted design of new materials, said Kim, who has used machine learning to explore applications of quantum many-body physics. With its ability to analyze vast amounts of data generated during materials research – such as large, real-time datasets from the Cornell High Energy Synchrotron Source (CHESS) – AI opens that possibility. The goal of NSF AI-MI is to harness the rising tide of materials data, using AI to enable scientists to develop new materials based on prediction while also developing trustworthy AI and deepening our fundamental understanding of AI.
“We have reached a key moment in the development of artificial intelligence and materials research when the integration of the two fields will lead to powerful new mechanisms for discovery and development,” said Kavita Bala, Cornell provost, who in 2021 as dean of Cornell Bowers led the launch of the Cornell AI Initiative. “Building on a foundation we’ve already established here at Cornell, AI-MI collaborators are poised to find solutions in quantum computing, sustainable energy and other realms at a faster pace and in greater depth than ever before.”
AI-MI, which will be one of Cornell Research and Innovation Centers and Institutes, fits perfectly into the vision of the Cornell AI Initiative, said Thorsten Joachims, interim dean of Cornell Bowers and director of the Cornell AI Initiative.
“This new institute brings together Cornell’s leadership in artificial intelligence with its excellence in the physical sciences,” Joachims said, “and it showcases how collaborations around AI can develop transformative new approaches in many disciplines across campus.”
NSF has funded national AI Research Institutes since 2020, establishing centers for AI in astrophysics, food systems, engaged learning and many other areas. AI-MI, one of five new NSF institutes established in 2025 with a $100 million investment, adds to these the first and only institute dedicated to materials research.
“Artificial intelligence is key to strengthening our workforce and boosting U.S. competitiveness,” said Brian Stone, performing the duties of the NSF director. “Through the National AI Research Institutes, we are turning cutting-edge ideas and research into real-world solutions and preparing Americans to lead in the technologies and jobs of the future.”
AI-MI plans to create the AI Materials Science Ecosystem (AIMS-EC), an open, cloud-based portal that couples a science-ready large language model (LLM) with targeted data streams, including experimental measurements, simulations, images and scientific papers.
Building on the arXiv research paper repository hosted by Cornell and operated by Cornell Tech, AIMS-EC will facilitate work on many fronts, including: discovering two-dimensional moiré structures with properties suitable for robust quantum bits; designing new superconductors; and developing molecules for removal of microplastics from the environment.
This significant AI system at Cornell will accelerate the work of researchers around the world, said AI-MI co-PI Kilian Weinberger, professor of computer science at Cornell Bowers and an expert in machine learning and its applications in computer vision, natural language processing and scientific discovery.
“Most LLM-based AI systems, like ChatGPT, are amazing in many ways, but very general. If you get too specific and ask rare, targeted questions, the models become unreliable and can start to ‘hallucinate,’” he said. “We can address these issues through specialization but also through new AI architectures – in particular, some that separate knowledge from language competency.”
Working with AI experts will accelerate the discovery of quantum materials, said Darrell Schlom, Tisch University Professor in the Department of Materials Science and Engineering (Cornell Engineering). An expert in oxide molecular beam epitaxy-based thin film growth, he will be working with image recognition experts Kim, Weinberger and Jennifer Sun, assistant professor of computer science (Cornell Bowers), to gain AI feedback from the real-time diffraction data his group collects during synthesis of thin films of quantum materials.
“These real-time data are so overwhelmingly rich in information that it typically takes a human about a year to recognize what parts provide important clues on how to improve the synthesis,” Schlom said. “We anticipate that with the help of AI – in the hands of the experts – the time needed to optimize the synthesis of a desired quantum material will be sped up by 10 to 100 times.”
In addition to research, AI-MI will implement an educational program. The institute will partner with high schools and universities, including Cornell partners K-12 Initiative at Cornell Tech and CHESS, to help prepare students of all levels for careers at the intersection of AI and the physical sciences.
.jpg)
