Design generative educational tools for stage-aware, hybrid authoring workflows combining automation with direct manipulation¶

Strategy · All strategies
Evidence · no claims cited

Description¶

The article recommends that generative educational tools support multiple forms of control aligned to authoring stages rather than optimizing for a single paradigm. As the authors state, "generative tools should support multiple forms of control that align with exploration, structuring, and finalization, allowing users to move fluidly between stages as their goals change". This follows directly from teachers' stage-dependent preferences and their requests for hybrid workflows combining prompt-based generation with object-level editing.

Design Implications¶

Context¶

Requirements¶

• Systems must expose inspectable structure and support precise, localized correction so that verification and correction remain low-cost.

Constraints¶

• The recommendation derives from a single-session study with primary mathematics teachers; the article notes interaction preferences may differ for other grade levels or subjects.

Target Learners¶

• primary mathematics teachers creating instructional visuals

Target Learning Goals¶

• producing numerically correct, classroom-ready visual materials for mathematics instruction

Affordances¶

• Control Timing Design Space Ai Visual Authoring

Related Strategies¶

Examples¶

-

Key Sources¶

• Zhengxu Li, Junling Wang, and April Yi Wang. (2026). When Should Teachers Control AI Generation for Mathematics Visuals? arXiv preprint. https://arxiv.org/abs/2605.10672