The Human Element of Data and AI

Teachers today know that every student learns differently. But the way teachers are taught inevitably impacts their own understanding of education and learning. For Gahyun Callie Sung, what she feels was missing from her own educational journey has become the focal point of her research and practice.

“I think I’ve always been interested in ways to understand and document the struggles of students who are falling between the cracks,” says Sung, Ph.D.'24, a recent graduate in Human Development, Learning, and Teaching. Growing up in South Korea, Sung was intimately aware of the educational pressures that many students there feel.

“Korea has some of the unhappiest students in the world, unfortunately,” says Sung. “And I wasn’t really a stranger to it.”

Though doing well academically, Sung says she struggled “socially and emotionally in high school,” and it made her think a lot about the goals of education, how learning is designed, and how we measure success in teaching. After earning a bachelor’s degree in English and master’s in educational/instructional technology at Seoul National University, Sung applied to Ph.D. programs where she could study the data educators use to better understand struggling students.

She learned about the research Associate Professor Bertrand Schneider was conducting at Harvard’s Learning, Innovation, and Technology (LIT) Lab and applied to the Ed School. Sung called the LIT Lab “one of the best things to happen to me at HGSE” and explained how Schneider’s research, in particular, interested her.

“He was doing work in multimodal learning analytics, which is to use different sources of data to understand student states,” Sung says. “That’s a very specific field inside the bigger field of learning analytics.”

Six years later, Sung has earned both her Ph.D. as well as her master’s in computer science at Harvard School of Engineering and Applied Sciences (SEAS). This spring, she presented her dissertation at HGSE. Working with Schneider and the LIT Lab, Sung applied her own learning experiences to her research and practice and saw herself transform from a curious student to a researcher adding her own work to the field.

“I think there’s been a slow transition over the years where you start off as someone who is a learner and who consumes knowledge. At some point, you start to realize you’re actually producing something,” says Sung, noting that it took time to focus her field of study at HGSE. “Sometimes people come here with very specific interests. I wasn’t one of those people, I had to do a lot of narrowing down and a lot of boiling down to get to a point where I’m at a dissertation and a career.”

Sung had an assigned seat in the lab space, what she calls a “luxury in Cambridge,” that allowed her to build relationships with other students and researchers who knew where to find her developing her work, including the three studies that became her dissertation.

“Callie has strived during her time at HGSE, because her love for research has made her take full advantage of what the community has to offer,” says Schneider, who calls her a “central figure” in the LIT Lab. “She is just a very kind, patient, supportive, but also passionate and driven person. She has helped other Ph.D. students develop their research agendas, and always brings a positive outlook to any research.”

Sung’s three-part dissertation aims to “paint a different picture with student data” by studying a particularly vulnerable population: newcomers to science, technology, engineering, and mathematics (STEM). Sung drew upon her own experiences studying computer science, which she said was an oftentimes “lonely experience” that sees students get discouraged and, in some cases, walk away from the field entirely.

Sung’s work aims to address the “leaky pipeline” of STEM using data and feedback that bridges the gap between identifying students who are struggling and the creative interventions that make meaningful impact on students.

“I think a lot of learning analytics applications and even a lot of research right now is focused on end performance and grades and achievement. It’s all tied to accountability culture in schools right now,” Sung says. “But that’s a limited view of the student, and sometimes people don’t even realize it because of the power of the data, but they’re painting a very narrow, datafied student and then they’re not really capturing their human struggles. So how do you use data to do that?”

Her first two dissertation studies tracked the student wellbeing of novice programmers learning the Python coding language for the first time, using data from wristbands and webcams focusing on their facial expressions and overall levels of stress. The third study of her dissertation used a generative artificial intelligence chatbot to help craft automated student feedback in a makerspace environment.

While Sung readily admits the scope of that final study was limited — focusing on a specific learning environment and particularly high grade of student — the results showed that AI technology can be applied in a practical, impactful way to help struggling students.

“The motivating question for me here is when you deal with such different types of data and different ways of straining or creating a picture of these data you start to realize there’s a lot of different ways to paint very different narratives of students,” says Sung. “I want to break open the black box, so to speak, and have people understand what’s going on and see how that’s going to be taken into consideration when you’re putting AI into the classroom.”

Sung’s research will continue after graduation from HGSE this spring, as she will join the University of Iowa as an assistant professor at their Learning Sciences and Educational Psychology program. Sung expects to continue the same kind of research in Iowa, which is welcome news to her dissertation adviser.

“Her motivation comes from a personal and deep interest in helping struggling students strive,” Schneider says. “I strongly believe that the field needs more people like Callie.”