Dr. Tina Lasisi has been using AI since the very beginning, and her vision for how it can support research in EEB is different from the standard Silicon Valley approach. “There's a level of automation that most science is not built for,” Lasisi said. “It reminds me of people talking about the industrial revolution. Not everything can be made in a factory. There are machines that can knit but no machines that can crochet. There’s a lot of crocheting in science.”
Like crochet, science still requires tools. Lasisi, assistant professor in EEB and Anthropology, has partnered with Dr. Lin Ma, assistant professor in Computer Science and Engineering, to create MiDuRAG, Michigan Dual-layer Reasoning and Analysis Graphs. One of 37 funded projects in the inaugural year of the Strategic Initiative Fund, the project aims to provide researchers with a workshop full of focused, specific AI tools, not a fully automated factory. Keeping the human at the center of the process of scientific discovery, MiDuRAG provides targeted assistance to speed along the tedious tasks, not the higher-level reasoning.
“Right now you have two extremes: people who want to completely throw away AI because it's unreliable, makes mistakes, and doesn't explain its reasoning, and people who want to tell the robots to do go do a thing, and they're going to do it, and we're going to take their word for it,” Lasisi said. By creating a middle ground, Lasisi and Ma are helping to repurpose AI tools from their origin in computer science into disciplines with data that play by different rules.
Interdisciplinary collaboration is key to MiDuRAG’s approach. When Lasisi gave a set of her own data to a student in Ma’s lab to experiment with, the student felt the AI tools were doing a fairly good job with the analysis. As a domain expert, however, Lasisi quickly saw that its conclusions didn’t really answer the question, and that it had simply found a correlation with no real point. “That tells us something about the way that knowledge is structured and captured in different fields,” she said, “And it explains why AI might give us something very different than what experts might give us. That's exciting.”
Lasisi hopes to make science more efficient, not to fundamentally change the process. “There's no progress bar where we will be 50% done with science, and one day 100% done with all of the knowledge,” she said. “The dimensions of what we're doing are constantly expanding. With every revolution that we have had of new technologies, it's not like we've come closer to the end. It's that we have discovered new questions, that the dimensions of knowledge expansion and production have changed.”
MiDuRAG contributes to the Life-Changing Education and Advanced Technology impact areas of the Strategic Initiative Fund. President Domenico Grasso announced the $1 billion, 10-year Strategic Initiative Fund in November 2025, seeking ideas with the potential to create lasting, transformational change across our campus community and society. Congratulations to Tina Lasisi and Lin Ma for their incredible work in bringing this research to EEB!
