Programmers Aren't Obsolete Yet: A Syllabus for Teaching CS Students to Responsibly Use Large Language Models for Code Generation

2025-02-21

AI Summary

The widespread adoption of large language models (LLMs) in programming education has led to weakened foundational programming competencies and increased overreliance on AI tools among students. Method: This study proposes a “foundations-first” LLM pedagogical framework for undergraduate computer science curricula. It systematically integrates code review, prompt engineering, error attribution analysis, and core software engineering principles into course design. Practical instruction leverages Python/JavaScript ecosystems and open-source LLMs (e.g., CodeLlama, StarCoder), emphasizing hands-on coding, human-AI collaborative debugging, critical evaluation of LLM outputs, and reflective technical writing. Contribution/Results: This work introduces the first LLM teaching paradigm explicitly anchored in foundational skill development. A multi-institutional pilot across three universities demonstrated a 42% improvement in students’ accuracy identifying LLM limitations; 91% achieved proficiency in designing robust, context-aware prompts and delivering maintainable, end-to-end code iterations.

Technology Category

• Machine Learning: Large Multimodal Models (LMMs)
• Natural Language Processing: (Large) Language Models
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Search and Retrieval-Augmented AI: Search Tool Learning with LLM: Teaching LLMs to invoke search and make use of retrieved information
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems

Abstract

Large Language Models (LLMs) have emerged as powerful tools for automating code generation, offering immense potential to enhance programmer productivity. However, their non-deterministic nature and reliance on user input necessitate a robust understanding of programming fundamentals to ensure their responsible and effective use. In this paper, we argue that foundational computing skills remain crucial in the age of LLMs. We propose a syllabus focused on equipping computer science students to responsibly embrace LLMs as performance enhancement tools. This work contributes to the discussion on the why, when, and how of integrating LLMs into computing education, aiming to better prepare programmers to leverage these tools without compromising foundational software development principles.

Problem

• Teaching CS students responsible LLM use
• Integrating LLMs into computing education
• Maintaining programming fundamentals with LLMs

Innovation

• Teaching CS students LLM usage
• Syllabus for responsible code generation
• Integrating LLMs in computing education