Using Adapted Conjecture Maps to Foster Interdisciplinary Collaboration Between Learning Scientists and Novice AI-Ed Researchers

Abstract

As the rate of progress in artificial intelligence (AI) research accelerates, learning scientists increasingly engage in deep collaboration with AI researchers to design learning tools. In entering into these interdisciplinary collaborations for the first time, AI researchers bring with them epistemic and practical commitments to design and research that differ significantly from the learning sciences. In this paper, we demonstrate a case study where learning scientists and AI researchers use an adapted form of a conjecture map as a boundary object to improve outcomes of design-based research. We highlight two moves, “zooming in” and “looking across”, that support boundary crossing moments that weave connections between technical decisions and learning processes. We then show how these moves lead to conversations around mitigating unintended consequences of designs based in artificial intelligence.