Collaborative learning with block-based programming: investigating human-centered artificial intelligence in education

Abstract

In this article, we investigate human-centered artificial intelligence (HCAI) in an educational context where pupils used block-based programming in small groups to solve tasks given by the teacher. We used a design-based research approach in which we, together with the teachers, created a maker space for explorative science learning and organised teaching interventions wherein the pupils met online three hours a week for 16 weeks for an entire school year. Due to COVID-19, data were collected through Zoom, with collaborative learning situations captured through screen sharing and online communication using webcams. We employed three data analysis techniques: interaction analysis, visual artifact analysis, and thematic analysis. We developed an analytical framework for integration using thematic coding that combined concepts from computer-supported collaborative learning (CSCL) and domain-oriented design environments. We report the following findings: 1) Three types of rules between design units were identified with visual artifact analysis: latent, generic, and domain-specific rules; 2) two types of CSCL artifacts (technology and discussions) were intertwined and developed in parallel, along with a computer-based scaffolding scenario that offloads domain-specific scaffolding from humans to computers.