Research map / Large language models / Agents, code and math
A Theory of Adaptive Scaffolding for LLM-Based Pedagogical Agents
Clayton Cohn, Surya Rayala, Namrata Srivastava, Joyce Horn Fonteles, Shruti Jain, Xinying Luo, Divya Mereddy, Naveeduddin Mohammed, Gautam Biswas
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add chain-of-thought explanations to few-shot examples (ablation) | Inquizzitor GPT-4o grader · FA2 scoring | weighted kappa: 93.58 -> 96.03 | Improved |
| add chain-of-thought explanations to few-shot examples (ablation) | Inquizzitor GPT-4o grader · FA3 scoring | weighted kappa: 94.45 -> 96.12 | Improved |
| add two few-shot examples without explanations to I/O prompt (ablation) | Inquizzitor GPT-4o grader · FA4 scoring | weighted kappa: 89.73 -> 80.69 | Hurt |
| add few-shot examples without explanations to I/O prompt (ablation) | Inquizzitor GPT-4o grader · FA2 scoring | weighted kappa: 91.28 -> 93.58 | Improved |
| add chain-of-thought explanations to few-shot examples (ablation) | Inquizzitor GPT-4o grader · FA4 scoring | weighted kappa: 80.69 -> 84.91 | Improved |
| add active learning prompt revisions after CoT (ablation) | Inquizzitor GPT-4o grader · FA4 scoring | weighted kappa: 84.91 -> 90.61 | Improved |
| full prompt pipeline versus plain I/O prompt (ablation) | Inquizzitor GPT-4o grader · FA4 scoring | weighted kappa: 89.73 -> 90.61 | No clear effect |
| add CoT to few-shot examples (ablation) | Inquizzitor GPT-4o grader · FA4 scoring | micro-F1: 78.00 -> 78.00 | No clear effect |
| add ICL examples to I/O prompt (ablation) | Inquizzitor GPT-4o grader · FA4 scoring | micro-F1: 82.00 -> 78.00 | Hurt |
| prompt agent with goal-setting construct instructions | Inquizzitor · student conversations | goal-setting faithfulness: 28.59 -> 18.36 (FA2 to FA4); unfaithfulness 53.38 -> 56.87 | Hurt |
| prompt agent with ZPD and self-efficacy construct instructions | Inquizzitor · student conversations | ZPD faithfulness: 59.26 (FA2), 65.88 (FA3), 62.20 (FA4) | Improved |
| instruct agent to resist score change requests | Inquizzitor · student conversations | score changes: fewer than 1% of cases | Improved |
Closest papers by meaning
- Distilling LLM Agent into Small Models with Retrieval and Code Tools (NeurIPS 2025 · similarity 0.88)
- Scaling Equitable Reflection Assessment in Education via Large Language Models and Role-Based Feedback Agents (AAAI 2026 · similarity 0.88)
- SocraticLM: Exploring Socratic Personalized Teaching with Large Language Models (NeurIPS 2024 · similarity 0.87)
- Unveiling the Learning Mind of Language Models: A Cognitive Framework and Empirical Study (NeurIPS 2025 · similarity 0.86)
- CogMath: Assessing LLMs' Authentic Mathematical Ability from a Human Cognitive Perspective (ICML 2025 · similarity 0.86)
- Developing LLM-Powered Trustworthy Agents for Personalized Learning Support (AAAI 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
