National Science Foundation AI Institute for Engaged Learning (EngageAI)

• In Event: Cross-Cutting Research Across Three NSF (National Science Foundation )-Funded National AI Institutes on Education

Abstract

Objectives or purposes

The mission of the NSF AI Institute for Engaged Learning (EngageAI) is to produce transformative advances in STEM learning and teaching with AI-driven narrative-centered learning environments. The EngageAI Institute is guided by a vision of learning that supports and extends the capabilities of students and teachers with AI. The Institute conducts research on narrative-centered learning environment technologies, embodied conversational agents, and multimodal learning analytics to create deeply engaging collaborative story-based learning experiences. Driven by a learner-centered vision of AI-augmented learning, the Institute develops learning environments that create narrative-centered learning experiences designed to promote student engagement in both formal and informal learning settings. Our research is informed by a nexus of stakeholders to ensure that AI-empowered learning technologies are ethically designed and prioritize diversity, equity, and inclusion.

Perspective(s) or theoretical framework

Our formulation of narrative-centered learning is guided by problem-based learning, anchored instruction, cognitive apprenticeship, learning-by-teaching and other approaches to collaborative inquiry that ground learning in meaningful tasks.

Methods, techniques, or modes of inquiry

We take a design-based approach to research that involves theory-based conjectures, participatory design, research in real-world educational settings, and iterative evidence-based refinement of both theory and design. In particular, we are testing conjectures about how AI can augment narrative-centered learning environments and its impact on learning and engagement.

Data sources, evidence, objects, or materials

Our research employs qualitative and quantitative methods in tandem with machine learning to investigate collaborative inquiry in narrative-centered learning environments. For example, understanding students’ epistemic dialogue contributions during collaborative game-based learning can provide a lens into group dynamics and insight into how dialogues develop. This understanding can highlight how meaning-making occurs by specifically focusing on the development and evolution of knowledge in group discourse.

Results and/or substantiated conclusions or warrants for arguments/point of view

We developed a large language model-based computational approach for recognizing epistemic dialogue acts from small groups of middle school students while interacting within a collaborative game-based learning environment. The model outperforms competitive alternatives in student dialogue act classification.

Scientific or scholarly significance of the study or work

Our research to uncover how learners in a range of contexts engage in collaborative inquiry is informing the design of narrative-centered learning environments that will enable the automatic generation of narratives that support learners at the individual, small group, and whole class level. To this end, we have created a multimodal learning analytics framework to support students, researchers, teachers, and informal STEM educators to greatly expand their awareness and pedagogical capacity through innovations in natural language processing, computer vision, and machine learning methods. Rich streams of multimodal data derived from students’ conversations, gaze, facial expressions, gesture, and posture as they interact with each other, with teachers, and with embodied conversational agents, support powerful learning analytics. Because our research has a specific emphasis on examining the relationship between learning and engagement, we have designed the multimodal learning analytics pipeline to support explorations of disciplinary engagement, collaborative engagement, and socioemotional engagement, as well as their relationship to learning processes and outcomes.

Author

• James C. Lester, North Carolina State University