Research & Science

AI tutors should help more, then step back

Constant help can steal the thinking that builds judgment. This paper argues for tutoring that responds to what the learner does. Support rises when needed, then fades so the student still makes key decisions.

Beyond Prompting
Read original source

What the source reports

Xinmeng Hou and coauthors describe an AI tutor that changes its support based on learner behavior. The system can add help when a learner struggles. It can also remove help when the learner is ready to decide. The goal is to preserve the mental work involved in planning, checking, and adjusting one’s own thinking. The paper argues that useful support depends on when it appears and when it steps back.

Original source

Title
Adaptive Scaffolding Needs Contingency: An AI Tutor That Escalates and Fades on What the Learner Does
Author
Xinmeng Hou, Yuxuan Weng, Chin Hsien Yeh, Ding Lin Lee, Lishan Zheng, Fang Li, Wuqi Wang, Yang Liu
Publication
arXiv
Date
Saturday, September 19, 2026