Advancing the Design and Implementation of Artificial Intelligence in Education through Continuous Improvement

Abstract

With the rapid rise of Artificial Intelligence in Education (AIEd), multiple stakeholders are questioning AI’s capability to make fair and trustworthy decisions that improve teaching and learning. We suspect that unfair and unreliable outcomes might stem from lack of systematic collaboration between the developers of AIEd systems and the educators tasked with their implementation. In a profession that is underresourced, teachers don’t merely need technology-centered solutions. Rather they and the students they serve need useful tools that work in culturally and socially complex instructional environments. In this article, the authors argue that FATE in AIEd-related issues must be addressed as the system evolves with users and the local context. This requires supporting the development of users’ ownership over AIEd systems that is needed to adapt them to their local contexts. It is in this process of gaining ownership that significant issues related to the FATE of AIEd systems present themselves. Inspired by continuous improvement approaches, we propose that the pursuit of FATE of AIEd lays broadly in: (a) promoting systematic inquiry and collaboration between educators, developers, and researchers; (b) exploring, through collaborative efforts, how on-the-ground realities influence the implementation of AIEd; and (c) using variation as an opportunity to learn how to make a system work reliably and across contexts. The authors conclude by discussing the implications of continuous improvement for research, development, and practice of AIEd.