Search for: All records

Research indicates that teachers play an active and important role in classrooms with AI tutors. Yet, our scientific understanding of the way teacher practices around AI tutors mediate student learning is far from complete. In this paper, we investigate spatiotemporal factors of student-teacher interactions by analyzing student engagement and learning with an AI tutor ahead of teacher visits (defined as episodes of a teacher being in close physical proximity to a student) and immediately following teacher visits. To conduct such integrated, temporal analysis around the moments when teachers visit students, we collect fine-grained, time-synchronized data on teacher positions in the physical classroom and student interactions with the AI tutor. Our case study in a K12 math classroom with a veteran math teacher provides some indications on factors that might affect a teacher’s decision to allocate their limited classroom time to their students and what effects these interactions have on students. For instance, teacher visits were associated more with students’ in-the-moment behavioral indicators (e.g., idleness) than a broader, static measure of student needs such as low prior knowledge. While teacher visits were often associated with positive changes in student behavior afterward (e.g., decreased idleness), there could be a potential mismatch between students visited by the teacher and who may have needed it more at that time (e.g., students who were disengaged for much longer). Overall, our findings indicate that teacher visits may yield immediate benefits for students but also that it is challenging for teachers to meet all needs - suggesting the need for better tool support. 

Past research shows that teacher noticing matters for student learning, but little is known about the effects of AI-based tools designed to augment teachers’ attention and sensemaking. In this paper, we investigate three multimodal measures of teacher noticing (i.e., gaze, deep dive into learning analytics in a teacher tool, and visits to individual students), gleaned from a mixed reality teacher awareness tool across ten classrooms. Our analysis suggests that of the three noticing measures, deep dive exhibited the largest association with learning gains when adjusting for students’ prior knowledge and tutor interactions. This finding may indicate that teachers identified students most in need based on the deep dive analytics and offered them support. We discuss how these multimodal measures can make the constraints and effects of teacher noticing in human-AI partnered classrooms visible. 

Constructing effective and well-balanced learning groups is important for collaborative learning. Past research explored how group formation policies affect learners’ behaviors and performance. With the different classroom contexts, many group formation policies work in theory, yet their feasibility is rarely investigated in authentic class sessions. In the current work, we define feasibility as the ratio of students being able to find available partners that satisfy a given group formation policy. Informed by user-centered research in K-12 classrooms, we simulated pairing policies on historical data from an intelligent tutoring system (ITS), a process we refer to as SimPairing. As part of the process for designing a pairing orchestration tool, this study contributes insights into the feasibility of four dynamic pairing policies, and how the feasibility varies depending on parameters in the pairing policies or different classes. We found that on average, dynamically pairing students based on their in-the-moment wheel-spinning status can pair most struggling students, even with moderate constraints of restricted pairings. In addition, we found there is a trade-off between the required knowledge heterogeneity and policy feasibility. Furthermore, the feasibility of pairing policies can vary across different classes, suggesting a need for customization regarding pairing policies. 

Orchestration tools may support K-12 teachers in facilitating student learning, especially when designed to address classroom stakeholders’ needs. Our previous work revealed a need for human-AI shared control when dynamically pairing students for collaborative learning in the classroom, but offered limited guidance on the role each agent should take. In this study, we designed storyboards for scenarios where teachers, students and AI co-orchestrate dynamic pairing when using AI-based adaptive math software for individual and collaborative learning. We surveyed 54 math teachers on their co-orchestration preferences. We found that teachers would like to share control with the AI to lessen their orchestration load. As well, they would like to have the AI propose student pairs with explanations, and identify risky proposed pairings. However, teachers are hesitant to let the AI auto-pair students even if they are busy, and are less inclined to let AI override teacher-proposed pairing. Our study contributes to teachers’ needs, preference, and boundaries for how they want to share the task and control of student pairing with the AI and students, and design implications in human-AI co-orchestration tools. 

Educational AI (AIEd) systems are increasingly designed and evaluated with an awareness of the hybrid nature of adaptivity in real-world educational settings. In practice, beyond being a property of AIEd systems alone, adaptivity is often jointly enacted by AI systems and human facilitators (e.g., teachers or peers). Despite much recent research activity, theoretical and conceptual guidance for the design of such human–AI systems remains limited. In this paper we explore how adaptivity may be shared across AIEd systems and the various human stakeholders who work with them. Based on a comparison of prior frameworks, which tend to examine adaptivity in AIEd systems or human coaches separately, we first synthesize a set of dimensions general enough to capture human–AI hybrid adaptivity. Using these dimensions, we then present a conceptual framework to map distinct ways in which humans and AIEd systems can augment each other’s abilities. Through examples, we illustrate how this framework can be used to characterize prior work and envision new possibilities for human–AI hybrid approaches in education.