Insights from Testing the Community Builder AI Partner (CoBi) in Classrooms

• In Event: Next-Generation Educational AI Research From Three NSF-Funded National AI Institutes

Abstract

Objectives
The NSF National AI Institute for Student-AI Teaming (iSAT) is developing AI Partners – intelligent technology which collaborates with teams of students and teachers to promote development of AI literacy, STEM competencies, and 21st century skills. Here, we present the Community Builder AI Partner (CoBi) that focuses on the relationship dimension of collaboration. CoBi helps students to co-negotiate classroom agreements that reflect their idealized interactions with each other. CoBi then uses AI technologies to look for evidence of these agreements during small group collaborative learning (Figure 1). The results are aggregated across student groups (to protect student privacy) and visualized to promote class-level reflections about the extent to which students are realizing their agreements (Figure 2).

Theoretical framework
CoBi was motivated by the responsible innovation framework (Stilgoe et al., 2013). Accordingly, CoBi was co-designed via a series of workshops with diverse youth where expressed a desire for AI to help them build strong communities in class (see Figure 2). CoBi is centered on four community agreements that are derived from the Open Sci Ed K-12 science materials (http://www.openscied.org/) aligned with national standards: being respectful, being equitable, being committed to community, and moving our thinking forward.

Methods and Data sources
Here, we focus on analyzing the results of a pilot deployment of CoBi in several middle school classrooms between March 2023 and May 2024. The primary data source analyzed here are thematically-coded observations and field-notes taken by trained researchers while CoBi was used in classrooms.

Results
Overall, students showed active engagement during the initial brainstorming and consensus-building discussion to establish co-negotiate agreements. Turning to the CoBi feedback visualizations (i.e., animated tree), as soon as growth of the tree was spotted, students reacted with positive astonishment and excitement (e.g., gaping students, laughter, and verbal praise) to the dynamic CoBi visualizations.

Scholarly significance
CoBi offers a novel approach by making the relationship dimension of collaboration explicit and provides mechanisms for students to collaboratively reflect on their small group interactions with their teacher.

Figures

Fig. 1. CoBi supports establishing, expression, and (repeated) revisiting of community agreements.

Fig. 2. Left: Students discuss CoBi results with the teacher. Right: The CoBi Revisiting interface including classwide community agreements and visual feedback via an animated tree where leaves appear when CoBi finds evidence of student discourse whereas emergence of color-coded flowers appear when CoBi notices examples of community agreements in student discourse.

References
Authors. (2024). Co-design Partners as Transformative Learners: Imagining Ideal Technology for Schools by Centering Speculative Relationships. In Proceedings of the ACM CHI Conference on Human Factors in Computing Systems (CHI 2024) (pp. 566 (561-515)).

Stilgoe, J., Owen, R., & Macnaghten, P. (2013). Developing a framework for responsible innovation. Research policy, 42(9), 1568-1580.

Author

• Sidney K. D'Mello, University of Colorado - Boulder