Researchers at the University of California are using artificial intelligence (AI) to advance work in health, energy, agriculture, and meteorology. Their efforts come as public concern about AI grows; according to recent surveys, half of Americans are more worried than excited about how AI is changing daily life. Still, many recognize its potential benefits, especially in science.
UC scientists are approaching AI with caution. They are exploring questions such as whether AI can be trusted with high-stakes decisions and how to ensure that human judgment remains central to scientific progress.
In meteorology, Ashesh Chattopadhyay, an applied mathematician at UC Santa Cruz, is working to improve weather forecasting with AI. Traditional forecasting relies on massive amounts of data and computational power. Chattopadhyay’s team developed FourCastNet in collaboration with NVIDIA, CalTech, and Rice University. This AI model uses decades of weather data to make predictions much faster and with less computing power than conventional methods. The European Centre for Medium-Range Weather Forecasting has started using FourCastNet and similar tools from Google and Huawei.
However, research from Chattopadhyay’s group shows that current AI models struggle with extreme events outside their training data. “AI works great for day-to-day weather over, say, Houston,” Chattopadhyay says. “But what about when Houston is facing something that’s never been seen in recorded history, like Hurricane Harvey?” When tested on unprecedented storms by removing major hurricanes from its training set, FourCastNet underestimated storm intensity. “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,” he adds.
The team is now working on integrating climate trend algorithms into short-term forecasting pipelines to improve predictions of rare events.
In agriculture, Alireza Pourreza at UC Davis leads the Digital Agriculture Laboratory developing Leaf Monitor—an AI system that scans leaves to assess plant health in real time. Using a handheld spectrometer and spectral analysis tied to nutrient signatures like nitrogen or potassium levels, Leaf Monitor provides immediate feedback compared to traditional lab testing which can take weeks. Pourreza explains: “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.” He believes tools like Leaf Monitor can help farmers make better decisions for crop management.
In healthcare, Hannah Milch at UCLA studies the use of AI in breast cancer screening. While more mammograms are being read with help from AI each year—and several algorithms have FDA clearance—Milch notes there is limited evidence showing improved patient outcomes in clinical settings. “These tools advertise that they’ll help us catch more cancers… but the evidence behind those claims is based largely on someone reading a couple of hundred mammograms in a lab setting,” she says.
Milch’s 2025 study found that Transpara—an AI tool—spotted about 30 percent of breast cancers missed by radiologists during routine screenings among nearly 185,000 mammograms reviewed between 2010 and 2019. She says: “It’s encouraging to see that if that radiologist were in that situation again…that cancer may have been caught five, six, eight months earlier.” Milch is now co-leading the PRISM trial—the largest randomized prospective study of AI in breast cancer screening—to determine if outcomes improve when radiologists use these tools.
For energy research, Mohammad Javad Abdolhosseini Qomi at UC Irvine is leading Geophysicist.ai—a project funded by a $6 million grant from the UC Office of the President—to use AI for safer geothermal energy development. Advances in drilling technology now allow engineers to reach deep underground heat sources previously inaccessible except through fracking techniques—which have raised concerns due to increased seismic activity observed in places like Oklahoma between 2008 and 2019 as fracking expanded throughout the state (https://www.usgs.gov/news/featured-story/induced-earthquakes). Qomi’s project aims to combine large language models with physics-based simulations and real-world drilling data so engineers can better predict conditions underground and select safe sites for geothermal projects.
Across these fields—from weather prediction and farming practices to medical diagnostics and clean energy—UC researchers emphasize careful evaluation before widespread adoption of new technologies. As they develop new applications for artificial intelligence within their disciplines, they continue prioritizing safety and public benefit alongside technological advancement.
