A new AI-powered tool developed by researchers from the University of Houston and the Environmental Molecular Sciences Laboratory (EMSL) can predict how fluids and dissolved chemicals move through rocks and other porous materials underground. This method combines artificial intelligence with established physical and chemical laws to provide faster and more reliable predictions. The team’s findings were recently published in the journal Transport in Porous Media. Modeling subsurface environments is particularly complex because fluids and chemicals travel through intricate networks of pores and fractures, and researchers often have limited information about what is happening underground. The EMSL–University of Houston team uses a type of AI called a physics-informed neural network (PINN). Unlike traditional machine learning, which relies mostly on data, PINNs are trained using both data and the scientific principles that govern fluid flow and chemical reactions underground. As the model learns, its predictions are continuously checked against these laws. If a prediction violates the rules of physics or chemistry, the model adjusts its calculations to align with both the data and the underlying science. A crucial aspect of this approach is ensuring that the model’s results align with basic scientific principles. For example, if a model predicts the amount of a mineral or chemical in a rock formation, that value should never be negative. However, small computational errors can sometimes lead to unrealistic results, especially when dealing with minerals that exist in very small quantities and are hard to detect. The team’s framework is designed to keep predictions within realistic limits while still adhering to the physical and chemical rules that define the system. This new modeling technique has potential applications in various areas related to critical mineral recovery, such as in situ mining, biomining, acid mine drainage, and recovery from waste piles or produced waters. In these scenarios, understanding how fluids should be injected, how long they should move through materials, and what conditions could maximize mineral recovery is essential. By predicting how fluids, dissolved minerals, and chemical reactions interact underground, the framework can help researchers test fluid injection and recovery strategies virtually before applying them in the field, potentially reducing costs and improving mineral yields. The team’s framework represents an important first step in developing physics-informed machine learning tools for critical mineral recovery. So far, the framework has been tested on simplified problems that capture key processes involved in mineral recovery. Future versions could incorporate more complex chemistry, biology, or microbial processes to create more realistic models of underground systems. This research ties into EMSL’s broader efforts to integrate modeling, experiments, and data. Through EMSL Community Science Campaigns, researchers are exploring new methods that use experimental platforms such as laboratory chips, imaging data, and porous materials to better understand how minerals move and react. These experiments help validate models, while the models can guide future experiments. The long-term goal is to develop AI-enabled digital twins for critical mineral and leaching operations. A digital twin is a computer model that updates itself with new data, creating a feedback loop between experiments, field observations, and predictions. While this study is an early step in that direction, it provides a foundation for adaptive tools that could one day help guide mineral recovery in real time.