In the spring of 2025, the University of Tennessee launched the AI TechX initiative to spur artificial intelligence (AI) research and innovation while increasing the AI preparedness of the state of Tennessee’s workforce. Over 30 teams led by UT faculty and their industry partners applied for the initial round of AI TechX seed grants—but selected just nine winning projects, each of which was awarded $60,000.
One of the inaugural awards went to Mathworks Professor Michela Taufer and Research Assistant Professor Kin Ng for their proposed collaboration with International Business Machines Corporation (IBM).
This collaboration will address a gap in the current space of large language model (LLM) technology. There has recently been a great deal of investment and progress in training LLMs, but the output applications of such models—called prediction or inference—have lagged.
“IBM is advocating that it’s time to look at that second stage,” said Taufer. “However, while they have advanced AI models and the knowledge in inference, they would like to see these models applied to scientific applications.”
IBM has a longstanding collaboration with Taufer, Ng, and other researchers in the Min H. Kao Department of Electrical Engineering and Computer Science working to integrate AI into cyberinfrastructure. When the AI TechX applications opened, Taufer and Ng’s collaborators at IBM suggested using the opportunity to apply the company’s AI models and inference tech to applications currently under study at UT.
“The goal of our project is to develop mini applications that we can optimize using IBM’s large foundation models,” Ng said. “This project will accelerate multiple areas of science, including earth science and bioinformatics, where applications are usually not optimized at the inference level.”
For example, Taufer and Ng’s teams have been working to assess wildfire risk using machine learning, which involves processing high-resolution maps of different variables across a broad area.
“That kind of research can bring a lot of computational bottlenecks on cyberinfrastructure,” Ng explained. “We’re trying to optimize on those applications such that the inference is faster, more efficient, and more effective.”
Workforce Development and Visibility
UT’s Global Computing Lab (GCLab), headed by Taufer, has a strong history of graduating students who are valuable to the nation’s computer science workforce; in fact, GCLab’s collaborating organizations frequently recruit from the lab’s ranks. UT alumna Paula Olaya (PhD/EECS ’24), who worked with IBM during her time with the GCLab, was recruited by IBM after graduating (though she is not working on this project).
The AI TechX seed funding continues a fruitful collaboration with IBM that Taufer and Ng are confident will lead to further opportunities for EECS students, especially those working directly on the project, like EECS master’s student Jason “Chandler” Weeks.
“At UT, we are performing workforce development in an interdisciplinary environment with collaborators in industry and at other universities,” Taufer said. “That prepares our students to lead impactful, interdisciplinary work as they take the next steps in their careers.”
Research connections with high-profile industry partners also help the science itself gain recognition.
“This is a unique opportunity to make a tangible scientific impact and help that impact gain the recognition it deserves,” Ng said. “Academic research can be overlooked, but collaboration with an industry leader like IBM brings visibility to the innovative work being done.”
Contact
Izzie Gall (egall4@utk.edu)