Reflective Dialogue or Prompt Refinement? Effects of Tutor Scaffolding on Students' Independent LLM Use for Programming

2026-07-03

AI Summary

This study investigates how intelligent tutoring strategies based on large language models (LLMs) can enhance students’ ability to independently and effectively use LLMs in programming education. Comparing Socratic guidance (SG) with prompt refinement (PR), the research employs educational experimentation and behavioral analysis to provide the first empirical evidence that, although SG is perceived as less efficient, it more effectively fosters understanding-driven prompting strategies among learners. In subsequent unconstrained interactions with LLMs, students in the SG condition demonstrated significantly higher learning gains and superior prompting behaviors compared to those in the PR condition. These findings highlight the critical role of guided dialogue in cultivating long-term autonomous learning capabilities with AI tools.

Abstract

While Large Language Models (LLMs) can provide personalized support in learning, several studies have raised concerns regarding their use in education. Importantly, learning depends on how students engage with LLMs. This study examined how two types of LLM-based tutors shape students' prompting practices, learning, and subsequent LLM-use: a Socratic-Guidance (SG) tutor, which structures interaction through dialogic questioning, and a Prompt-Refinement (PR) tutor that guides the formulation of effective prompts. We conducted a two-phase study in a graduate-level mobile robotics course: 66 students used either the SG or PR tutor during a 6-week intervention, followed by 52 students using an unconstrained LLM during a 3-week course project. Results show that while the SG- and PR tutors led to similar task performance and prompting patterns during guided use, they differ in learning outcomes and later LLM-use. SG-students, relative to PR-student, achieved higher learning gains in later sessions, and were more likely to adopt understanding-driven prompting strategies, which are predictive of higher understanding, when using an unconstrained LLM. Although learners perceived the SG tutor as less efficient, the findings suggest that Socratic guidance supports the development of students' capacity to learn with LLMs over time, highlighting its importance for LLM tutor design.

Problem

• Large Language Models
• Socratic Guidance
• Prompt Refinement
• Programming Education
• Student Learning

Innovation

• Socratic Guidance
• Prompt Refinement
• Large Language Models
• Tutor Scaffolding
• Understanding-driven Prompting