Facilitating Instructors-LLM Collaboration for Problem Design in Introductory Programming Classrooms
Project Overview
The document explores the integration of generative AI, particularly Large Language Models (LLMs) like ChatGPT, in educational settings, specifically for designing programming problems in introductory courses. It emphasizes the collaborative potential between instructors and LLMs, showcasing how these models can assist in generating innovative programming tasks, providing constructive feedback, and clarifying common misconceptions among students. By employing a participatory design approach, the study developed an instructor-authoring tool that leverages LLM capabilities to enhance the efficiency and effectiveness of creating educational activities. The findings indicate a range of perceptions regarding the usefulness, efficiency, and creativity of LLMs in instructional design, highlighting a critical need for human oversight to ensure quality and relevance. Overall, the document underscores both the promise and challenges of utilizing generative AI in education, advocating for a balanced approach that combines AI capabilities with teacher expertise.
Key Applications
INSIGHT Classroom Assistant
Context: Introductory programming courses at universities, targeting instructors and students.
Implementation: Instructors use the INSIGHT authoring tool to collaborate with LLMs for generating programming problems, solutions, and feedback.
Outcomes: Improved efficiency and effectiveness in designing programming problems, better coverage of learning objectives, and enhanced instructor workflows.
Challenges: Concerns about the accuracy of LLM-generated content, over-reliance on generic templates, and the need for human oversight to ensure the quality of generated materials.
Implementation Barriers
Technological
LLMs may generate inaccurate or generic content that lacks depth and creativity.
Proposed Solutions: Instructors should verify LLM-generated content, use structured prompts to guide the generation process, and implement a mediatory layer to refine any feedback before providing it to students.
Human Factor
Instructors may lack familiarity with using LLMs effectively, leading to inefficiencies in the unguided approach.
Proposed Solutions: Provide training and resources on best practices for utilizing LLMs in instructional design.
Project Team
Muntasir Hoq
Researcher
Jessica Vandenberg
Researcher
Shuyin Jiao
Researcher
Seung Lee
Researcher
Bradford Mott
Researcher
Narges Norouzi
Researcher
James Lester
Researcher
Bita Akram
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Muntasir Hoq, Jessica Vandenberg, Shuyin Jiao, Seung Lee, Bradford Mott, Narges Norouzi, James Lester, Bita Akram
Source Publication: View Original PaperLink opens in a new window
Project Contact: Dr. Jianhua Yang
LLM Model Version: gpt-4o-mini-2024-07-18
Analysis Provider: Openai