Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

MazeMate: An LLM-Powered Chatbot to Support Computational Thinking in Gamified Programming Learning

Chenyu Hou, Hua Yu, Gaoxia Zhu, John Derek Anas, Jiao Liu, Yew Soon Ong

TL;DR

MazeMate integrates an LLM-powered chatbot into a gamified 3D Maze programming environment to scaffold computational thinking across maze solving and maze design. By coupling real-time maze data with path-finding backends and staged prompts, it aims to provide adaptive, context-aware CT support while mitigating AI hallucinations. In a classroom study with 247 undergraduates, MazeMate yielded moderate perceived usefulness, with clearer benefits for solving tasks; thematic analysis confirmed CT facilitation in decomposition, pattern recognition, abstraction, and algorithmic thinking, though designers noted mismatches and fabricated solutions especially in design tasks. The work demonstrates the potential of AI-assisted scaffolding in authentic classrooms and outlines concrete design directions to enhance grounding, transparency, and engagement for CT development.

Abstract

Computational Thinking (CT) is a foundational problem-solving skill, and gamified programming environments are a widely adopted approach to cultivating it. While large language models (LLMs) provide on-demand programming support, current applications rarely foster CT development. We present MazeMate, an LLM-powered chatbot embedded in a 3D Maze programming game, designed to deliver adaptive, context-sensitive scaffolds aligned with CT processes in maze solving and maze design. We report on the first classroom implementation with 247 undergraduates. Students rated MazeMate as moderately helpful, with higher perceived usefulness for maze solving than for maze design. Thematic analysis confirmed support for CT processes such as decomposition, abstraction, and algorithmic thinking, while also revealing limitations in supporting maze design, including mismatched suggestions and fabricated algorithmic solutions. These findings demonstrate the potential of LLM-based scaffolding to support CT and underscore directions for design refinement to enhance MazeMate usability in authentic classrooms.

MazeMate: An LLM-Powered Chatbot to Support Computational Thinking in Gamified Programming Learning

TL;DR

MazeMate integrates an LLM-powered chatbot into a gamified 3D Maze programming environment to scaffold computational thinking across maze solving and maze design. By coupling real-time maze data with path-finding backends and staged prompts, it aims to provide adaptive, context-aware CT support while mitigating AI hallucinations. In a classroom study with 247 undergraduates, MazeMate yielded moderate perceived usefulness, with clearer benefits for solving tasks; thematic analysis confirmed CT facilitation in decomposition, pattern recognition, abstraction, and algorithmic thinking, though designers noted mismatches and fabricated solutions especially in design tasks. The work demonstrates the potential of AI-assisted scaffolding in authentic classrooms and outlines concrete design directions to enhance grounding, transparency, and engagement for CT development.

Abstract

Computational Thinking (CT) is a foundational problem-solving skill, and gamified programming environments are a widely adopted approach to cultivating it. While large language models (LLMs) provide on-demand programming support, current applications rarely foster CT development. We present MazeMate, an LLM-powered chatbot embedded in a 3D Maze programming game, designed to deliver adaptive, context-sensitive scaffolds aligned with CT processes in maze solving and maze design. We report on the first classroom implementation with 247 undergraduates. Students rated MazeMate as moderately helpful, with higher perceived usefulness for maze solving than for maze design. Thematic analysis confirmed support for CT processes such as decomposition, abstraction, and algorithmic thinking, while also revealing limitations in supporting maze design, including mismatched suggestions and fabricated algorithmic solutions. These findings demonstrate the potential of LLM-based scaffolding to support CT and underscore directions for design refinement to enhance MazeMate usability in authentic classrooms.

Paper Structure

This paper contains 26 sections, 8 figures, 1 table.

Table of Contents

  1. Introduction
  2. Related Work
  3. Computational Thinking and Programming Learning
  4. Scaffolding for Computational Thinking
  5. BACKGROUND: 3D MAZE
  6. 3D Maze Introduction and Activities
  7. Prior Implementation of 3D Maze with Generative AI
  8. Implementation Procedures and Data Collection
  9. Identified challenges from the previous round of implementation of the 3D Maze
  10. MazeMate Designs
  11. Technical Details of MazeMate
  12. MazeMate Provides Contextualized and Adaptive Support
  13. MazeMate Provides Accurate Solutions through Path Finding Algorithm
  14. Prompting for Progressive Scaffolding of MazeMate
  15. Alignment of MazeMate Scaffolds with Computational Thinking Processes
  16. ...and 11 more sections

Figures (8)

  • Figure 1: Collaborative with group members in the 3D maze system.
  • Figure 2: Maze design process in the 3D maze system.
  • Figure 3: The overview of the interaction process between students and MazeMate.
  • Figure 4: Maze design process with the help of MazeMate in 3D maze system.
  • Figure 5: Maze solving process with the help of MazeMate in 3D maze system.
  • ...and 3 more figures