Skip to main content Skip to navigation

Prompts First, Finally

Project Overview

Generative AI (GenAI) and large language models (LLMs) are revolutionizing computer science education by allowing students to interact with programming through natural language, thereby enhancing accessibility and engagement. The document advocates for a 'Prompts First' approach, which emphasizes the critical role of teaching prompt engineering to maximize the benefits of GenAI while addressing concerns related to academic integrity and dependency on AI tools. It draws parallels between the evolution of programming languages toward greater abstraction and the emergence of GenAI as a significant advancement in this trajectory. The authors call for innovative pedagogical strategies to incorporate GenAI into educational curricula effectively, ensuring that students not only harness its capabilities but also develop a deeper understanding of programming concepts. Overall, the findings suggest that with careful integration, GenAI can enhance learning outcomes and prepare students for future challenges in technology and programming.

Key Applications

Generative AI for Programming Education

Context: Introductory programming courses for students, utilizing LLM-generated explanations and problem-solving support.

Implementation: Employing generative AI tools for natural language programming and prompt engineering, along with research and development of educational materials that leverage LLM outputs for code explanations.

Outcomes: ['Increased engagement and understanding of programming concepts without the need for perfect syntax.', 'Enhanced student learning through tailored explanations and accessible programming exercises.']

Challenges: ['Potential over-reliance on AI tools leading to an illusion of competence.', 'Issues with AI misinterpretation of prompts.', 'Variability in the quality of AI-generated content necessitating critical evaluation by students.']

Implementation Barriers

Educational Challenge

Lower-performing novice programmers may over-rely on Generative AI, disrupting their problem-solving skills.

Proposed Solutions: Educators should prioritize teaching debugging techniques and critical thinking to evaluate AI outputs.

Technical Challenge

Non-deterministic nature of AI models can lead to misunderstandings and incorrect code generation.

Proposed Solutions: Develop precise vocabulary for programming concepts and emphasize the importance of understanding the problem domain.

Project Team

Brent N. Reeves

Researcher

James Prather

Researcher

Paul Denny

Researcher

Juho Leinonen

Researcher

Stephen MacNeil

Researcher

Brett A. Becker

Researcher

Andrew Luxton-Reilly

Researcher

Contact Information

For information about the paper, please contact the authors.

Authors: Brent N. Reeves, James Prather, Paul Denny, Juho Leinonen, Stephen MacNeil, Brett A. Becker, Andrew Luxton-Reilly

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

Let us know you agree to cookies