STORY · FORSKNING_
TutorMoments: Do AI tutors know when to help and when to step back?
AI2 introduces TutorMoments, an evaluation framework based on real mathematics tutoring sessions to measure whether language models can balance when to support students and when to let them work independently. The research shows that models tend to over-help by providing too much support, and that this improves when the trade-off between help and challenge is made explicit in the model's instructions.
WHY IT MATTERS
This addresses a critical pedagogical challenge in AI tutoring: machine learning is optimized to be helpful, but effective teaching requires knowing when to hold back. The framework and dataset being released give the field a sharper tool for evaluating pedagogical AI behavior.
SOURCES
MACHINE-GENERATED SUMMARY This summary is written by machine from the sources below. We sort and explain — but we are a way into the field, not the final word. Check the source when something matters to you.