The rapid advancement of generative artificial intelligence has fundamentally transformed programming practices, enabling non-programmer designers to generate, modify, and deploy code through conversational prompts-an emerging paradigm referred to as ...
The rapid advancement of generative artificial intelligence has fundamentally transformed programming practices, enabling non-programmer designers to generate, modify, and deploy code through conversational prompts-an emerging paradigm referred to as vibe coding. Despite these technological advancements, designers continue to face psychological and technical barriers, including anxiety, reduced self-efficacy, limited technical understanding, and difficulties in problem solving, which constrain their effective use of AI-based code-generation tools.
This study aims to identify the structural characteristics of stage-based barriers experienced by designers in AI-assisted vibe coding and to propose practical guidelines for mitigating these barriers. The research comprises a preliminary interview study followed by three empirical studies. The preliminary interviews elicited barrier factors grounded in designers’ real-world AI coding experiences. In Study 1, structural equation modeling was employed to examine how psychological and technical barriers influence AI tool use and perceived development outcomes. Study 2 compared cognitive burden across the seven stages of the vibe coding process to identify critical bottleneck stages. Study 3 validated stage-specific coping strategies and the proposed guidelines through expert evaluation.
The results indicate that designers’ experiences in AI-assisted vibe coding are not merely driven by technical skill acquisition but represent a complex process in which psychological and technical factors affect tool use and performance outcomes through distinct pathways. In particular, significant bottlenecks were observed during the code generation, debugging, and organization–deployment stages, where cognitive fatigue and technical demands were most pronounced. The stage-based guidelines proposed in this study demonstrate potential for alleviating these burdens in the identified bottleneck stages.
Academically, this study contributes a designer-centered barrier structure model grounded in empirical user experiences rather than productivity-oriented assumptions. Practically, it offers actionable guidelines to support designers’ engagement in vibe coding and to enhance their performance in AI-assisted development workflows. Ultimately, this research provides a foundation for fostering a more inclusive and sustainable design-development ecosystem in the emerging era of designer-AI collaboration.
* Keywords: Vibe Coding, AI Code Generation, AI-Assisted Programming, Design-Development Convergence