Research
If you are interested in the list of publications from the lab, please go to the Publications page.
1. Using learning sciences principles to help humans interact effectively with AI systems
We use learning sciences expertise to design scaffolding, friction, and effective reflective interactions to help humans use AI systems effectively. Our research has discovered, among others:
- A simple warning that indicates that “AI can make mistakes” promotes users’ help-seeking behavior during problem solving (under review).
- A mindfulness intervention (mindful language + breathing exercises) in an adaptive learning system supports teens reduce math anxiety while enhancing math performance (CHI’25 LBW) [link]
- Armenian kids envision that their trust in virtual agents is shaped not only by visual appearance but also by empathetic and communication factors (IDC’25). [link]
2. Modeling and supporting self-regulated, strategic learning within AI/adaptive systems
We co-design learning activities and technologies with school teachers and students to understand and support meaningful, strategic learning activities (e.g., self-regulated learning, informed choice making). Our research has discovered, among others:
- Students show different choice-making behaviors during problem solving depending on their prior knowledge, and some demonstrate distinctive “strategic disengagement” (npj Science of Learning 2025) [link]
- Students whose choices were supported through self-reflection on their own choice-making behaviors in intelligent tutoring software chose to use an instructional aid less frequently yet more strategically, and learned greater domain knowledge and skills (ISLS’23) [link]
- Students, when their choices on the use of visual representations were guided with recommendations, seemed to learn and apply strategic choice-making decisions in a transfer environment where their choices are no longer guided. (IDC LDT’24) [link]
3. Socio-technical inquiry on AI through participatory research
We want to deeply understand how stakeholders perceive and feel about the use of AI tools in various complex environments, what they prefer, and why. We design AI systems based on this deep understanding through participatory research. Our research has found, among others:
- K-12 Teachers’ and students’ preferences towards classroom AI are distinctively different, causing a perspective gap. In particular, teacher-student social relationshipsaffect their views (e.g., trust) towards AI in the multi-stakeholder classroom environment (CSCW’26)
- Student agency when learning with AI in the classroom is a multi-dimensional concept, consisting of several different aspects of agency in making decisions around data, content, orchestration, and feedback/help, rather than a one-dimensional concept. (AIED’25: Best Paper Nominee) [link]
- The analysis of student interview data through the lens of “perspective taking,” we found that students compromise their own needs and preferences in using AI because they recognize other stakeholders’ roles within the classroom. (AIED’25) [link]
4. Cognitive understanding of human learning and adaptive instruction to support it
We design effective instructional techniques using adaptive technologies to support students’ cognitive learning in STEM domains, and use participatory and learning analytics approaches to understand how learning happens. Our research has discovered, among others:
- A scaffolded self-explanation activity with diagrams in intelligent tutoring software supports effective and efficient algebra learning (CogSci’21) [pdf]
- Students who learn with Intelligent tutoring software that gives feedback on learner effort do not outperform students who receive feedback on learner performance in math (CogSci’24) [link]
- A playful learning experience with drag-and-drop interaction and gamification features in intelligent tutoring software supports engaging and effective algebra learning during remote schooling (ECTEL’21) [link]
5. Researcher-practitioner ecosystem
We envision the future of learning research where we will have an ecosystem that systematically allows researchers and practitioners (e.g., educators, parents, and children) to jointly conduct learning research that informs both learning theories and real-world practice. We reflect on our practice regularly and aim to develop such an ecosystem.
6. Design and evaluation of learning analytics tools
We work with educators and students to co-design and evaluate reflection tools and prompts (e.g., learning analytics dashboards) to support teaching and learning.
7. Open Educational Resources (OER)
We work with policy researchers, education researchers, and practitioners to develop policies, strategies, and materials that promote equitable access to education.
