EAGER: Co-Designing a Cognitive Teaching Assistant to Support Evidence-Based Instruction in Open-Ended Learning Environments

Collaboration is a key ingredient of STEM professions and integrating computing is also crucial to science learning today. Thus, developing educational support technology that helps build a STEM-ready workforce that is equipped to collaborate productively and use computational models and tools effectively has broad ramifications for the future of a STEM workforce in the US. At a time when we are witnessing exponential growth in applications of AI and machine learning, this proposal leverages those advances to develop a novel AI technology-enhanced virtual teacher assistant that acts as a partner to teachers. In collaboration with Metro Nashville Public Schools, the team will work with five middle school science teachers and 200 students to co-design and deploy the virtual teaching assistant. This system provides on-demand feedback to teachers as they engage with their students in an integrated science, computing, and engineering curriculum that focuses on redesigning their schoolyard to minimize water runoff and cost while maximizing accessibility for all. The project will investigate key factors that promote student learning, including teachers’ instructional practices and curricular adaptations, and their interactions with the virtual teaching assistant. Our research will establish the impact of the AI-based teacher assistant on student outcomes and demonstrate how we can leverage today’s technology-enhanced classrooms for more successful, equitable, and sustainable introduction of real-world, problem-based STEM learning. This project will adopt a design-based research approach to develop a novel AI technology-enhanced cognitive teacher assistant that aids teachers in noticing, reflecting, and developing of evidence-based pedagogical responses, enriches classroom interactivity, and helps students progress in their learning and problem-solving tasks. We will study how well our technology identifies key classroom interactions, engages teachers in students’ learning and problem solving in a computational modeling-based science curriculum, and establishes its impact on key student outcomes. We will investigate key factors that promote these outcomes, including teachers’ instructional practices and curricular adaptations, and their interactions with the cognitive teaching assistant to support technology-enhanced noticing, reflection, and response practices. Results will be analyzed using traditional quantitative and qualitative methods as well as multi-modal learning analytics to understand how a diverse set of middle school teachers leverage the cognitive teacher assistant technology to orchestrate an open-ended, problem-based learning curriculum in their science classrooms. This work will 1) contribute to our understanding of how virtual teacher support agents can augment STEM+C teacher noticing, reflection, and response practices, and 2) produce guidelines to help the design of innovative teacher + AI agent partnerships that facilitate STEM+C integration. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.