Teacher-AI Co-Design Patterns for Responsible Classroom Automation

This mock summary covers a conference paper about teacher-AI co-design. The authors begin with a practical concern: many AI education tools are introduced as efficiency systems, but teachers evaluate them through a different lens. Teachers ask whether the tool respects their professional judgment, whether it understands classroom context, whether it creates new work, and whether it gives students fair support. To answer these concerns, the paper reports a design study with teachers, school leaders, and learning technologists. The study maps recurring design patterns for responsible classroom automation.

The first pattern is “suggest, do not silently decide.” Teachers preferred AI systems that prepare options, draft feedback, or surface evidence, while leaving meaningful decisions visible and reversible. The second pattern is “show the educational reason.” A recommendation should explain the learning objective, student evidence, and uncertainty behind it. The third pattern is “make automation adjustable.” Teachers wanted to set the level of automation based on task sensitivity. A spelling correction tool could act automatically, while assessment feedback or intervention grouping should require review. The fourth pattern is “protect relationship work.” Teachers were wary of AI messages that sounded supportive but lacked knowledge of the student. They preferred AI drafts that could be personalized by the teacher.

The paper contributes a framework that can guide product teams. It separates tasks into administrative support, instructional preparation, formative feedback, learner monitoring, and high-stakes judgment. For each category, it suggests design constraints, approval levels, and evidence requirements. The authors also introduce a lightweight evaluation rubric: agency, transparency, workload, fairness, and classroom fit. This rubric can be used before pilot deployment to identify risks that pure accuracy metrics may miss.

Findings show that teacher trust grows when tools make uncertainty visible and reduce routine effort without taking over the teacher's role. However, the authors warn that co-design should not become a one-off workshop. As AI capabilities and school policies change, teacher-facing tools require ongoing feedback channels, version history, and governance processes. Teachers also need professional learning that explains what a model can and cannot infer.

For AIEDHK, this paper matters because it positions teachers as co-designers of AI infrastructure. It suggests that Hong Kong's AIED hub strategy should not only track technical progress, but also document classroom workflow patterns, teacher concerns, and implementation playbooks. The paper is especially relevant for product development because it translates responsible AI principles into concrete interface and workflow choices.