Isabelle Freiling, a One-U RAI faculty fellow in communication, examines how people understand and interact with AI in the context of science communication, and why that matters for public health topics such as infectious diseases. “The problem is that not everyone is actually getting the same content displayed,” she said. “Without good access to that data, it's then really hard to do the research that we so urgently need.” She and her collaborators have identified a path forward.

Maggie French, a One-U RAI faculty fellow in physical therapy and athletic training, is using AI to improve post-stroke care. She’s developing a better way to assess patient mobility by using a computer vision model to analyze videos of patients’ movements. “This type of computer vision algorithm is actually very accurate, even in people who don't move like neurologically intact individuals,” she said. “We're now trying to bring this to the clinic.”

Pablo Maldonado, a One-U RAI postdoctoral fellow, is working to broaden access to mental health care services by developing an AI-powered tool accessible via smartphones. “Our goal here is to develop a low-cost, accessible, multimodal AI platform that learns human behavior and can identify psychiatric symptoms across different populations in Utah,” he said. It’s especially relevant in Utah, which has a shortage of mental health care providers and a higher rate of depression compared to the national average.

Vineet Pandey, a One-U RAI faculty fellow in computer science, is developing a touchscreen-based tool to remotely assess human motor performance—a boon for rural populations and movement-disorder patients who can’t easily travel to specialized clinics. “The possibilities are quite endless here—clinically valid data collection in remote settings,” he said of his tool. “It generates interpretable features that clinicians really appreciate.”

Mark Yandell, a One-U RAI faculty fellow in human genetics, co-leads Gene Kids, an AI-powered tool that analyzes the clinical records of children in intensive care units to determine within two days of admission if they’d benefit from genetic sequencing. When sequencing leads to definitive diagnoses, which it does for about half of patients, it’s “utterly transformative for care,” Yandell said.

Brian Codding, a One-U RAI faculty fellow in anthropology, is drawing from indigenous practices to develop an AI model that produces strategies to fight environmental crises such as the drying of the Great Salt Lake. Indigenous restoration efforts in Utah, he said, are already clearing invasive, water-hungry plant species, supporting native habitats for endemic species, and sending over 13,000 acre-feet of water back to the lake every year. “But how do we scale this up?” Codding asked. He’s using AI and data on natural capital (water, soil, species), economic output (revenue, tax benefits), land use (farm, conservation) and climate (precipitation, temperature) to figure that out. “We can train a model to help understand what the optimal scenario would be if we want to balance, for example, the trade-offs between economic output and restoring natural capital.”

Blake Vernon, a One-U RAI postdoctoral fellow, is leveraging AI to make agriculture more water efficient. The ratio of water supply to water demand is marginal at best, he said, and the biggest source of demand for water in the West is agriculture. We can help farmers “get a better sense of how water is and will be distributed in the soils that their crops have access to,” he explained. One major challenge: soil moisture is hard to measure across farm plots. So he’s integrating the sparse soil-moisture data that does exist from monitoring stations and satellite imagery with climate and terrain data. The end goal is a better tool to help farmers understand soil moisture and inform irrigation practices.

Di Wang, a One-U RAI postdoctoral fellow, designs AI agents to better understand and support disaster recovery. “We bring together a wide range of data,” he said, citing census records, property taxes, business patterns, satellite and street-level imagery, social media activities, and field research. “That gives us a multi-dimensional view of disaster recovery.” Wang and his collaborators plan to pilot their work with the Utah Division of Emergency Management.

Jim Agutter—faculty in multi-disciplinary design and a member of the committee that set goals for One-U RAI’s teaching and learning focus area—drew from his training as an architect to advocate that educators use AI as a “scaffold” to support learning rather than replace it. AI should be integrated into education in a way that encourages constructive, iterative, and human-centered learning processes, he said. “AI will be in the classroom. There's no question about that. The real question is, what will we choose to construct with it? What kind of architecture or learning will we leave behind for our students?”

Chenglu Li, a One-U RAI faculty fellow in educational psychology, manages ALTER-Math, a digital tool that is already helping thousands of middle schoolers learn math by teaching it to fictional, AI-powered peers. Learning by teaching can help students overcome educational hurdles tied to disengagement, lack of ownership and rote memorization—but it isn’t always realistic or scalable. “With the advancements of generative AI, we have a chance to reimagine this learning-by-teaching experience, where we can flip students' roles from students to the instructor,” he said. After reaching over 60,000 students in the first two years, the tool will reach 120,000 students in the next year, and Li and his team will continue to refine it through testing.

Ana Marasović, a One-U RAI faculty fellow in computer science, wants to harness her expertise in large language models to help faculty manage their many duties. “It is often asserted that AI can help us improve our time efficiency, the quality of our materials, the quality of our exams, improve our communications with students, help us manage our teaching assistants, and so on,” she said. However, faculty members are often excluded from the design of such tools. She is working to identify faculty needs, design AI prototypes, and co-create solutions with educators.

Jon Thomas—director of digital learning technologies for the university and a member of the committee that set goals for One-U RAI’s teaching and learning focus area— encouraged educators to strategically use AI tools to enhance learning when appropriate. “How could my students use OpenAI ChatGPT to accomplish a certain learning objective we're going through?” he asked. “What artifact could they produce that would demonstrate that learning objective?”