Open access
Human–AI Learning Evaluation
Published: 16 October 2026
Evaluating Socratic Dialogue Scaffolding by Fine-Tuned LLMs in Undergraduate Chemistry Tutoring
Elena Rostova, Douglas Jenkins, Tri Wulandari
1. Dept. of Chemistry Education, Universitas Tadulako; 2. Center for Human-AI Interaction; 3. Educational Technologies Institute.
Abstract
A double-blind randomized control trial comparing open-ended AI dialogue versus traditional static text feedback. The Socratic agent generated statistically higher conceptual transfer scores (d = 0.72) on complex stoichiometry exams.
Keywords:
Large Language Models, Socratic Tutoring, Chemistry Education, Dialogue Scaffolding, Generative AI
Publication Details
| Journal | Journal of Intelligent Pedagogy |
| Issue | Volume 1, Issue 1 (2026), pp. 13–24 |
| ISSN (Online) | 3025-8891 |
| Publisher | Global Innovation & Academic Press (GIA Press) |
| Peer Review | Double-blind peer reviewed in accordance with COPE ethical guidelines |
| License | Creative Commons Attribution 4.0 International (CC BY 4.0) |
How to Cite (APA 7th Edition)
Elena Rostova, Douglas Jenkins, Tri Wulandari (2026). Evaluating Socratic Dialogue Scaffolding by Fine-Tuned LLMs in Undergraduate Chemistry Tutoring. Journal of Intelligent Pedagogy, 1(1), 13–24. https://doi.org/10.59653/joip.v1i01.2027