Future Scientists Navigate AI‑Driven Research with MATRIX‑Uni

Allison McLevige brushed off concerns about her summer research project.

“There’s a lot of narrative these days about, like, ‘Oh, AI is going to take your job,’” said McLevige, an undergraduate student in chemical engineering at the University of Illinois Urbana-Champaign.

But after spending several weeks conducting hands-on research with an AI institute, she had her own thoughts on the matter: “It’s a tool. It’s not a full replacement for a scientist.”

McLevige was part of the first cohort of undergraduate fellows assembled by the U.S. National Science Foundation’s Molecule Maker Lab Institute (NSF MMLI) for its inaugural MATRIX-Uni program. MATRIX, or MMLI Advancing Talent and Research through Interdisciplinary eXcellence in AI and Chemistry, was designed to bridge the gap between science and emerging technology. MATRIX-Uni puts that mission into practice by pairing undergraduate students with NSF MMLI graduate mentors and faculty members for an immersive summer research experience.

With a background in chemistry, McLevige came to the program with limited exposure to computational methods, so when MATRIX-Uni announced her project placement, evaluating the performance of large language models (LLMs), she was taken aback. “I was like, they did read my application, right? Like, they know I’m not a CS student,” she recalled thinking upon being assigned to a computer science lab.

But soon enough, that unfamiliar space became an opportunity to develop new technical skills in machine learning, LLMs, and neural networks and learn how to apply them to chemistry.

Another fellow, Mariia Martova, approached that disciplinary divide from the opposite direction. As a computer science major, Martova was comfortable with computational methods but landed a project in biochemistry, specifically engineering membrane sugar transporters. As a community college student transferring to the University of Illinois Urbana-Champaign, Martova came to MATRIX-Uni intrigued by the opportunity to try something new.

“One of my main goals was to see how I can apply my computer science skills to a different industry, instead of what people mostly think about when they start computer science, which is software engineering,” said Martova.

Assigning a computer science student to biochemistry and a chemistry student to computer science might look like a mix-up at first glance. In MATRIX-Uni, crossing those disciplinary boundaries is the name of the game.

NSF MMLI graduate mentor Chetan Prasad sees that intersection firsthand in his own research. “When you work on a data-driven project, you have to collect a lot of data points,” he said. “There are so many technical things that you need to know, and you need a student who is willing to do a bit of CS, a bit of engineering, and a bit of chemistry.”

Prasad found that student in Jelena Cheng, a senior in materials science and engineering at Case Western Reserve University. Working under Prasad’s mentorship, Cheng helped develop machine learning models to predict polymer solubility, gaining experience with coding and machine learning that complemented her materials laboratory experience. By working across disciplines, she broadened her technical skills while getting a firsthand look at the demands of graduate-level research.

That experience also helped answer a larger question. Cheng came to MATRIX-Uni wanting to know whether she would still enjoy research long-term. Even after some long days in the lab, the answer was clear.

“No matter how hard I work the first day, I still come in the lab with happiness. Like, I still enjoy the work,” Cheng said. “So I think, yeah, it has definitely confirmed my decision for going to grad school.”

The boundaries MATRIX-Uni helped fellows navigate were not only disciplinary. For Soohwan Kim, an undergraduate in computer science at the University of Illinois Urbana-Champaign, graduate research itself initially seemed distant and intimidating. “At the beginning, it just seemed really daunting to, like, first step into the lab.… I remember the first day I wasn’t sure if I was allowed to come in,” Kim confessed.

Several weeks of working alongside graduate students and researchers brought clarity and reassurance. What grounded his perceptions was “seeing how normal everyone is once you get to know them, because as an undergrad, research roles and postdocs, grad students all seem so, like, far away,” Kim said. Those connections provided a valuable perspective as he considered his own next steps.

Mentors gained as much from the experience as the fellows they guided. By taking responsibility for mentoring a student and managing a research project, NSF MMLI graduate mentors gained practical experience in teaching, communication, problem-solving, and leadership. The role gave them an opportunity to guide another researcher through the challenges of a project while developing skills they will carry into future research and leadership roles.

At a time when conversations about AI often center on what technology might replace, from our jobs to our ingenuity, creativity, and problem-solving, the experiences of the MATRIX-Uni fellows offer hope. Here, working with AI meant learning more, not less. Fellows had to stretch beyond their own disciplines, develop new technical knowledge, evaluate the tools they were using, and apply their own scientific judgment to real research problems. Graduate mentors, meanwhile, developed skills distinctly human in nature and scope: leadership, communication, vision, and judgment.

Together, fellows and mentors contributed to active NSF MMLI research spanning drug discovery, enzyme engineering, protein design, and sustainable materials, while expanding their computational skills in AI and machine learning. These tools will undoubtedly advance research, but their impact will depend on the scientists who wield them. The next generation of scientists is evolving, and thanks to MATRIX-Uni, they have a sandbox.