Design AI tutoring behaviors that withhold solutions, prompt recall, and ask critical questions¶

Strategy · All strategies
Evidence · no claims cited

Description¶

A concrete implementable recipe drawn from what students themselves requested and from learning-science principles: feedback that withholds the solution, prompts active recall, offers alternative explanations, and guides learners through critical questioning and self-reflection. The paper specifies that scaffolding systems are ones "that know when to withhold an answer, ask a question, surface uncertainty, or present alternative perspectives." It cites Maike, a privacy-preserving Socratic educational chatbot, as an early illustration of this approach.

Design Implications¶

Context¶

Requirements¶

• Pedagogical alignment with active learning, constructivism, and scaffolding; managing cognitive load and promoting meta-cognition

Constraints¶

• Illustrated by an early chatbot design rather than validated at scale in this paper

Target Learners¶

• K-12 and higher-education learners using AI tutoring tools

Target Learning Goals¶

• effortful retrieval
• productive struggle
• critical thinking
• deep understanding

Affordances¶

• Self Reinforcing Harm Cycle Ai Education

Related Strategies¶

• Use predicted hint-taking likelihood and hint effects to adaptively decide whether to withhold or provide hints
• Hint Design
• Help Seeking
• Encouraging Students to Ask for Help
• Encouraging Help-Seeking Behavior

Examples¶

-

Key Sources¶

• Lucile Favero, Juan Antonio Pérez-Ortiz, Tanja Käser, Nuria Oliver. (2026). From Substitution to Scaffolding: Breaking the Self-Reinforcing Harm Cycle of AI in Education (and Beyond). arXiv preprint. https://arxiv.org/abs/2608.17451