Prompt Problems: A New Programming Exercise for the Generative AI Era
Project Overview
The document explores the integration of generative AI, particularly large language models (LLMs), into computer science education through an innovative pedagogical approach called 'Prompt Problems'. This method emphasizes teaching students the skill of crafting effective prompts for AI code generators, which is increasingly vital as LLMs become integral to programming. The authors introduce a web-based tool, 'Promptly', designed to facilitate this learning by hosting various prompt problems and enabling automated evaluation of the AI-generated code. Initial evaluations of this approach indicate that it significantly enhances student engagement and fosters computational thinking skills. However, the document also raises concerns regarding the potential over-reliance on AI tools and their implications for the development of traditional programming skills. Overall, the findings suggest that while generative AI can enrich the educational experience in computer science, careful consideration is needed to balance its benefits with the foundational skills necessary for future programmers.
Key Applications
Prompt Problems
Context: Computer science education, specifically in CS1 and CS2 courses.
Implementation: Deployed the Promptly tool in two introductory programming courses to teach students how to write prompts for LLMs to generate code.
Outcomes: Students reported enthusiasm and improved computational thinking skills, with many successfully solving problems in just a few attempts.
Challenges: Some students struggled with prompt crafting, leading to incorrect submissions, and there were concerns about over-reliance on AI for coding.
Implementation Barriers
Educational
Resistance from some students regarding the use of generative AI tools, fearing it may undermine their creativity and independent coding skills.
Proposed Solutions: Address concerns through discussions about the importance of traditional coding skills and the role of AI as an aid rather than a replacement.
Project Team
Paul Denny
Researcher
Juho Leinonen
Researcher
James Prather
Researcher
Andrew Luxton-Reilly
Researcher
Thezyrie Amarouche
Researcher
Brett A. Becker
Researcher
Brent N. Reeves
Researcher
Contact Information
For information about the paper, please contact the authors.
Authors: Paul Denny, Juho Leinonen, James Prather, Andrew Luxton-Reilly, Thezyrie Amarouche, Brett A. Becker, Brent N. Reeves
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