A new National Science Foundation-backed project at the University of Iowa College of Engineering is pioneering faster, more accurate, and efficient generative AI systems for creating images, text, and other content.

The two-year, $300,000 project is led by Soura Dasgupta, F. Wendell Miller Distinguished Professor and professor of electrical and computer engineering, with Raghuraman Mudumbai, ECE associate professor, as co-investigator.

Their research focuses on a popular type of artificial intelligence called “generative AI,” which uses a process known as diffusion to gradually build realistic content. Right now, the way these systems work is slow and not very accurate because they rely on complicated workarounds to reverse the diffusion process.

The Iowa team plans to develop a new, more direct method to reverse this process. If successful, their work could make AI tools faster, more efficient, and more reliable. It could also lead to major advances in how scientists understand and use these types of mathematical models.

This grant highlights the University of Iowa’s growing role in cutting-edge AI research and its commitment to solving complex problems with real-world impact.