From Black Holes to Large Language Models

The students earn a stipend for their work, and they will present their findings at the annual URSI Symposium on September 23. The keynote speaker, Mina Cikara ’03, Fredric Wertham Professor of Law and Psychiatry in Society at Harvard, participated in the URSI program while at Vassar.

Professor of Biology Justin Touchon, URSI Director for 2026–2027, noted that while cuts in government funding have caused many science research programs to be curtailed, URSI has remained fully funded through support from private foundations and individuals, including several Vassar alums.

“URSI has been going for 40 years now and is bigger than ever,” Touchon said. “A few of those projects were funded by external grants from the National Science Foundation and the National Institutes of Health, but the vast majority were funded directly through the College. These donations have enabled Vassar to give nearly 2,200 students research experiences. I am grateful for the dedication of our science faculty, which has made URSI possible. Without their willingness to devote two months of their summer to working with students, none of this would happen.”

We take a look at just a few of this year’s projects:

Learning Assessment in the Era of Large Language Models

Toby Fantl ’28, Yufan Wei ’28, and Associate Professor of Cognitive Science Joshua de Leeuw

Can large language models help us assess what students learn? That was the question Fantl and Wei attempted to answer during their 10-week stint as URSI researchers. Specifically, the students created an artificial intelligence platform that interviews students about concepts from an introductory statistics course. Their work is part of a nationwide program called CourseKata that develops educational materials, which was founded in 2017 by scientists from the University of California, Los Angeles, and California State University, Los Angeles.

Wei, a psychological science major from Beijing, China, encountered some bumps in the road when asking artificial intelligence bots for help with the project. “A lot of times, AI would give too much information when it posed questions to test students in the class,” she said. “I literally had to tell AI, ‘Don’t give any hints to students’ in the test questions.”

As the research came to a close, de Leeuw said the students had achieved most of their goals for the project and had learned new skills along the way. “We had to do a lot of coding with software we weren’t familiar with,” said Fantl, a cognitive science major from Columbus, Ohio.

Wei was especially pleased with this aspect of the work. “This wasn’t like taking a class,” she said. “It was more like doing a real job, and I think it will help me find employment in the future. There are a lot of practical applications to the work we were doing.”

Though working on a single project for an extended period differed substantially from other academic experiences at Vassar, Fantl approached his URSI project in ways that mirrored his coursework strategy. “I tried to simulate the experience I have during the school year by dividing up my tasks into categories that kind of simulated going to class,” he said.

The students completed much of the research they had planned to conduct, de Leeuw said, but would continue working with CourseKata during the fall semester. He added he had been impressed by how self-sufficient Fantl and Wei had been throughout their URSI experience. “I let them work out a lot of the kinks in their research and was impressed with how self-sufficient they were,” he said. “I didn’t have to micromanage them at all.”

Investigating ICE Data and Technology

Claire Harvey ’28, Em Sherer ’27, and Assistant Professor of Computer Science Logan Stapleton

In the 19 months since President Trump took office, the strategies and tactics of U.S. Immigration and Customs Enforcement (ICE) have generated widespread concern and fear among immigrants and organizations that support them. During their eight-week URSI project, Harvey and Sherer continued work they had begun with Stapleton during the fall of 2025, gathering and analyzing data about ICE’s activities.

“Our objective is to collect community-submitted information to help inform the public and raise awareness,” Stapleton said. “The biggest question we have is: ‘What can we do that isn’t just information gathering but finding ways to help people in the real world?’”

Sherer, an anthropology major from Chatham, Massachusetts, said he and Harvey had compiled records from official government sources as well as data from two Boston-based organizations that have been tracking ICE activities over the past year and a half. “The data indicate where ICE activities are happening and how and where they are arresting people,” Sherer said.

Some of that data is particularly disturbing, said Harvey, a mathematics major from Waterford, Virginia. She and Sherer found that an overwhelming majority of documented cases involved people who were detained but had never been convicted of a crime. They had been moved multiple times from one detention center to another, often in faraway locations, making it nearly impossible for their loved ones or immigration lawyers to find them. In one case, Harvey said, a man was moved 26 times. In another, an immigrant was deported to Honduras, a country where he had never lived.

Harvey said her URSI experience was both illuminating and inspiring. “We learned what it’s like to be part of a team and to figure out how the data we collected can help people,” she said. “It’s about knowing ICE’s tactics to be able to better respond.”

Stapleton was awed by the work Sherer and Harvey had done during the summer. “I’m very proud of them; it’s obvious they care about these atrocities,” they said. “Working with government data can be really messy, and they did incredible work. To see them grow into academic researchers was inspiring.”

In the face of the disturbing data, Sherer and Harvey said they were optimistic that their work and the efforts of community groups across the country were making inroads into tracking ICE activities. “In some ways, the situation is less chaotic than the general public thinks,” Harvey said. “It’s not hopeless. There’s a way through the storm.”

Computational Redistricting and Voting Rights Analysis

Angela Moon ’28, Cristian Castellanos ’27, and Assistant Professor of Mathematics and Statistics Daryl DeFord

Are recent federal court decisions posing a threat to effective representation for minority voters? That was the task undertaken by Moon and Castellanos during their 10-week URSI project this summer. The students gathered census data and employed statistical and mathematical tools to analyze how voting districts have been drawn and redrawn in the wake of these court decisions. Moon and Castellanos examined existing maps of election districts and generated new ones using these mathematical tools. And they concluded that some recent court decisions—including Louisiana v. Callais, a landmark Supreme Court case that was decided in April 2026—have diluted the provisions of the Voting Rights Act of 1965. In that case, the court rejected a new map of voting districts in Louisiana that included a district that would have favored minority voters, ruling that the map constituted “racial gerrymandering.”

Moon, a sociology and mathematics and statistics double major from Merrimack, New Hampshire, said her URSI experience had enabled her to employ both of her fields of study to analyze real-world problems. “I had never done an in-depth project like this before,” she said. “As I look back on how much I’ve learned, I realize I can use this knowledge [of statistical analysis] and my social science background to address some of these issues.”

Castellanos, a mathematics and statistics major from Poughkeepsie, said conducting research on a single issue for an extended period of time was something new in his Vassar experience. He added that he and Moon had also learned to code using a programming language called Julia to complete their tasks. “The mentorship we received from Professor DeFord was awesome,” he said, “and I learned how useful my math skills are in addressing problems in the real world.”

DeFord said the research carried out by Moon and Castellanos revealed that minority voting rights are being threatened and that these recent court decisions “could also lead to decreases in voting rights more broadly.”

“This project had many complexities, and the students were able to react to rapidly changing data,” he said. “There were court decisions being handed down as we were working on this project, and they were able to adapt accordingly. This wasn’t just a math exercise. It was an exercise that looked at what is happening in the country right now.”

Building Black Holes

Aleksandra Lada ’28 and Visiting Assistant Professor of Physics and Astronomy Peter Weck

When massive stars called supernovae are in the final stages of their existence, they often implode, creating what scientists have dubbed black holes, which have gravitational forces so strong that even light cannot escape them. Over the past 10 years, astronomers have observed real-world black holes, but little is known about what they actually are. In her research this summer, Aleksandra Lada ’28 “built” black holes by creating mathematical descriptions of them that are consistent with a quantum theory of gravity.

Lada, a physics and astronomy major from Stoughton, Massachusetts, said she chose this topic for her URSI research because she’s interested in studying celestial objects not by looking through a telescope but rather through mathematical computations. She said black holes fit this kind of research “because they create massive gravitational forces that have not been fully explained.”

Lada had to start the project by learning aspects of mathematical computation that she hadn’t learned in her math or astronomy classes at Vassar. “It was definitely a lot to digest at first,” she said. “If I hit a roadblock, I’d try to figure it out for 24 hours and then ask [Assistant Professor Weck] for help.”

Weck said he had been studying black holes for several years, “and this URSI project enabled us to dig more deeply into that research.” He said he had chosen Lada to collaborate with him this summer because he had taught her in one of his classes and worked with her in a student group he had formed, the Mathematical Physics Reading Club.

He said Lada’s work this summer had exceeded his expectations. “I’ve worked with PhD candidates on projects like this, and I was really impressed with how much she was absorbing as an undergraduate,” Weck said. He and Lada will continue to explore black holes during the coming academic year.

Lada said working on the project had helped to confirm her plans to conduct such research as a career at a college or university, “but I also want to teach so I can inspire young people to learn what I have learned,” she explained.

She added that Vassar had been exactly the right place to develop her post-Vassar plans. “Vassar’s liberal arts approach to learning helps you ask the right questions,” Lada said, “and having professors available to bounce ideas off has definitely helped me.”