University of California researchers are testing the capabilities and limits of artificial intelligence (AI) across a range of scientific fields, including meteorology, agriculture, health care, and energy. Their work aims to advance technology while maintaining a critical perspective on its reliability and societal impact.
In meteorology, Ashesh Chattopadhyay from UC Santa Cruz is working with partners such as NVIDIA, CalTech, Rice University, and the UC-managed Lawrence Berkeley National Laboratory to develop FourCastNet. This AI system predicts weather using past data rather than solving complex physics equations for each forecast. As a result, forecasts can be generated much faster and with significantly less computing power compared to traditional methods. The European Centre for Medium-Range Weather Forecasting has begun incorporating FourCastNet and similar AI tools into daily operations.
Despite these advances, Chattopadhyay’s research highlights current limitations. “AI works great for day-to-day weather over, say, Houston,” he says. “But what about when Houston is facing something that’s never been seen in recorded history, like Hurricane Harvey?” By training FourCastNet on historical data excluding the most extreme hurricanes and then testing it against conditions known to produce a Category 5 storm, the model underestimated storm intensity. “We found that it couldn’t really extrapolate beyond what it had seen in its training data,” Chattopadhyay explains. He adds: “Despite how good these models are with routine weather, getting the extremes right is still a problem. And those extremes are actually the thing scientists and forecasters care most about.” To address this gap, his team is integrating climate modeling algorithms into short-term forecasting pipelines.
In agriculture, Alireza Pourreza at UC Davis leads efforts to apply AI for crop monitoring through Leaf Monitor—a handheld spectrometer-based tool that analyzes light reflected by plant leaves to assess nutrient levels in real time. This system allows farmers to make timely decisions about water and fertilizer use without waiting weeks for lab results. Pourreza notes: “If we’re going to move towards sustainable, regenerative agriculture that has less impact on the environment, we need to be able to manage in a real-time, site-specific manner. Leaf Monitor is an example of a tool that can help us get there.”
For health care applications like breast cancer screening, Hannah Milch at UCLA has studied whether AI can enhance radiologist performance during mammogram interpretation. Although commercial AI algorithms have been approved by regulators for clinical use in recent years and adoption is growing rapidly among providers nationwide—including UCLA Health—evidence of improved patient outcomes remains limited so far. Milch observes: “These tools advertise that they’ll help us catch more cancers and make us radiologists do our jobs better, but the evidence behind those claims is based largely on someone reading a couple of hundred mammograms in a lab setting. That’s very different from the daily work of interpreting thousands of scans in the real world.” In one study covering nearly 185,000 mammograms over nine years with an algorithm called Transpara, her team found it identified roughly 30 percent of cancers missed by humans during screenings: “It’s encouraging to see that if that radiologist were in that situation again, and they had the AI support tool flagging that spot, that cancer may have been caught five, six, eight months earlier.” Milch now co-leads PRISM—the largest U.S.-based randomized trial assessing whether AI-assisted readings improve actual patient outcomes.
UC Irvine’s Mohammad Javad Abdolhosseini Qomi focuses on energy applications by leading Geophysicist.ai—a $6 million project funded by the UC Office of the President—to use AI for geothermal energy development. By combining large language models with physics-based simulations and field data from drilling sites across western states such as California and Nevada (where advanced geothermal resources exist), engineers aim to identify safe locations for enhanced geothermal systems capable of providing reliable clean energy around-the-clock.
The University of California emphasizes rigorous science conducted for public benefit as central to its approach toward emerging technologies like artificial intelligence—balancing innovation with caution about risks.
