ADVANCED MATERIALS & PROCESSES | JANUARY 2026 10 MACHINE LEARNING | AI AI POWERS CRITICAL MATERIALS SEARCH Scientists at the U.S. Geological Survey (USGS), along with the Defense Advanced Research Projects Agency (DARPA) and the Advanced Research Projects Agency–Energy (ARPA-E), launched a bold initiative called Critical Mineral Assessments with AI Support, or CriticalMAAS. The project helps strengthen collaborative research from the University of Minnesota and the Viterbi Information Sciences Institute (ISI) at the University of Southern California. The goal is to develop AI tools that can accelerate how the U.S. identifies and discovers mineral resources using a workflow known as critical mineral assessments. The project’s first challenge was extracting data from 100,000 historical maps, most of which exist only as scanned raster images. The raster format makes them difficult to analyze using AI, and manual digitization is slow and labor-intensive. Meanwhile, prior automated approaches struggle with the complex visual context and limited training data available for geological maps. So, the team developed Digmapper, a scalable system that automates the map digitization pipeline. When tested on more than 100 annotated maps from a DARPA- USGS dataset, Digmapper completed the process in under 25 minutes. Digitizing the historical maps speeds the process of assessing where critical minerals can be sourced, say scientists. The second major ISI-led effort, MinMod, addresses the need to unify mineral data from around the world into a searchable, AI-ready knowledge graph. “The goal of MinMod is to bring together everything we know about global mining sites,” says ISI researcher Craig Knoblock. So far, MinMod has processed tens of thousands of docu- ments on more than 679,000 sites and 190 different commodities. The team uses large language models and semantic technologies to extract, structure, and unify data into a machine-readable knowledge graph. This shared framework ensures different systems can access, analyze, and build on the same data with precision. As part of the broader CriticalMAAS initiative, the historical maps and MinMod knowledge graph serve as Example of extracting polygon features from geologic maps. Courtesy of W. Duan et al., 2025, doi.org/10.48550/ arxiv.2506.16006. the foundation for the program’s goal of using machine learning to predict undiscovered mineral deposits. isi.edu. MAGNETIC MATERIALS DATABASE VIA AI Researchers at the Uni- versity of New Hampshire are using artificial intelli- gence (AI) to accelerate discovery of new functional magnetic materials. They cre- ated the Northeast Materials Database of 67,573 mag- netic materials entries, including 25 previously unrecognized compounds that remain magnetic even at high temperatures. “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,” says Ph.D. student Suman Itani. Scientists know that many undiscovered magnetic compounds exist, but testing millions of element combinations in the lab is prohibitively time-consuming and expensive. Itani and his team built an AI system that can read scientific papers and extract those key experimental details. This data is then fed into computer models that identify whether a material is magnetic and how high a temperature it can withstand before losing its mag- netism. Going forward, the scientists believe the large language model behind this project could have widespread use beyond this database, particularly in higher education. For example, converting images to modern rich text format could be used to modernize library holdings. nemad.org.
RkJQdWJsaXNoZXIy MTYyMzk3NQ==