Design AI configurations that preserve generative work: feedback-only audiences, generate-then-compare, and calibrated teaching interactions¶

Strategy · All strategies
Evidence · 1 claim (1 for) · 1 study (1 theoretical), q1 · 0 of 1 report an effect size · 1 claim rests on one study

Description¶

The article recommends concrete configurations that keep generative work with the learner. "A particularly promising configuration has students self-explain their reasoning to an AI that then provides feedback on the completeness of the explanation, functioning as a responsive audience that identifies gaps without supplying the missing explanation itself". Related recommendations include having students generate their own summaries first and compare them with AI versions, and calibrating AI responsiveness in teaching-to-AI setups through strict pedagogical guardrails so anticipation, gap detection, and repair are preserved.

Design Implications¶

Context¶

Requirements¶

• AI responsiveness must be calibrated through strict pedagogical guardrails so the AI does not signal understanding when the student's explanation was unclear

Constraints¶

• If the AI is too capable, its responses may remove the feedback signal that makes teaching effective

Target Learners¶

• mathematics students using generative AI tools

Target Learning Goals¶

• self-explanation, summarizing, and teaching as generative strategies

Affordances¶

• Three Roles Ai Generative Learning Typology

Claims¶

• AI shifts the learner's role from producer to evaluator of mathematical meaning across multiple generative strategies [+W]

Related Strategies¶

• Counter-design practices for deliberate AI engagement
• Calibrate assistant support forms to preserve users' cognitive participation rather than making every interaction tutor-like
• Design AI learning assistants to make the implications of AI use visible rather than restricting assistance

Examples¶

-

Key Sources¶

• Hoyun Cho. (2026). Relocating the Locus of Generation: The AI-GLM Framework for Generative Learning in AI-Mediated Mathematics Education. Educational Psychology Review. https://doi.org/10.1007/s10648-026-10210-y