"Can you feel the vibes?": An exploration of novice programmer engagement with vibe coding

2025-12-02

AI Summary

This study investigates the learning mechanisms and pedagogical viability of “vibe coding”—natural language–driven software development—in novice programmers using AI-assisted programming. Through an interdisciplinary one-day hackathon, we employed behavioral observation, surveys, and semi-structured interviews to analyze collaborative patterns, tool usage, and learning outcomes. To our knowledge, this is the first empirical integration of vibe coding into formal computing education, demonstrating its efficacy in rapid prototyping and cross-disciplinary teamwork. We propose pedagogical scaffolds emphasizing divergent thinking guidance and critical evaluation of AI-generated outputs. A multi-model generative AI coding pipeline—augmented by human validation—was deployed to enhance code correctness and robustness. Participants delivered functional prototypes within nine hours, reporting significant gains in programming self-efficacy and prompt engineering proficiency. However, challenges emerged, including premature convergence of ideas, inconsistent code quality, and limited engagement with software engineering practices.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Cognitive Modeling & Cognitive Systems: Neural Spike Coding
• Humans and AI: Learning Human Values and Preferences

Application Category

• Economics, Online Markets and Human Computation: Research challenges in human and human-AI computation
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• User Modeling, Personalization and Recommendation: Practical large-scale studies of user experience

Abstract

Emerging alongside generative AI and the broader trend of AI-assisted coding, the term"vibe coding"refers to creating software via natural language prompts rather than direct code authorship. This approach promises to democratize software development, but its educational implications remain underexplored. This paper reports on a one-day educational hackathon investigating how novice programmers and mixed-experience teams engage with vibe coding. We organized an inclusive event at a Brazilian public university with 31 undergraduate participants from computing and non-computing disciplines, divided into nine teams. Through observations, an exit survey, and semi-structured interviews, we examined creative processes, tool usage patterns, collaboration dynamics, and learning outcomes. Findings reveal that vibe coding enabled rapid prototyping and cross-disciplinary collaboration, with participants developing prompt engineering skills and delivering functional demonstrations within time constraints. However, we observed premature convergence in ideation, uneven code quality requiring rework, and limited engagement with core software engineering practices. Teams adopted sophisticated workflows combining multiple AI tools in pipeline configurations, with human judgment remaining essential for critical refinement. The short format (9 hours) proved effective for confidence-building among newcomers while accommodating participants with limited availability. We conclude that vibe coding hackathons can serve as valuable low-stakes learning environments when coupled with explicit scaffolds for divergent thinking, critical evaluation of AI outputs, and realistic expectations about production quality.

Problem

• Explores how novice programmers use natural language AI coding
• Investigates educational impacts of AI-assisted programming on beginners
• Examines collaboration and learning outcomes in vibe coding hackathons

Innovation

• Vibe coding uses natural language prompts for software creation
• Teams combined multiple AI tools in pipeline workflows
• Human judgment remained essential for critical refinement