Vibe Coding in Practice: Building a Driving Simulator Without Expert Programming Skills
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper investigates the efficacy of "vibe coding," an AI-assisted development paradigm where users generate software through natural language prompts rather than manual coding. Motivated by the rise of Large Language Models (LLMs) and the potential to democratize software creation, the study addresses how this approach supports novice programmers in building complex systems. Specifically, it explores whether non-experts can create functional 3D driving simulators without prior coding experience, thereby lowering barriers to entry in computational tool development. The researchers conducted an exploratory case study involving a single participant with no programming background. Using the Cursor platform (an AI-integrated code editor) and the Three.js JavaScript library, the participant attempted to build a 3D driving simulator with procedurally generated environments. The methodology relied on an iterative, trial-and-error prompting process. The participant used natural language instructions, or "vibes," to request features such as vehicle physics, city realism, and camera controls. Two distinct development sequences were tested: one starting with minimal scenes and adding complexity incrementally, and another starting with high-level city prompts and refining details. The process involved continuous refinement, where the participant analyzed outputs, identified flaws, and adjusted prompts or pasted error messages to guide the LLM. The study produced two functional prototypes of the driving simulator. The results demonstrated that vibe coding significantly reduced traditional barriers to entry, allowing a non-expert to create interactive 3D environments without writing code. However, the quality and reliability of the outputs were heavily dependent on prompt specificity. Narrow, focused prompts yielded clearer and more stable results, while abstract or complex requests often produced incomplete code, logical flaws, or performance issues such as frame-rate drops. The findings indicate that achieving advanced interactivity and realism required multiple refinement cycles. Furthermore, the study revealed that effective vibe coding necessitates a new skill set centered on "prompt literacy," requiring users to possess technical reasoning, system analysis capabilities, and the ability to debug AI-generated code. The significance of this work lies in its demonstration that while LLMs can expand access to creative development, they do not eliminate the need for computational thinking. The authors conclude that software development is shifting from a hands-on technical task to a high-level design conversation, but developers must still understand logic, dependencies, and error identification. The paper highlights the limitations of current AI tools, noting that causal ambiguity remains due to the single-participant design and the interplay of various tools and hardware. Future research is recommended to include controlled comparisons, longitudinal studies, and standardized usability metrics to better evaluate the consistency, accessibility, and code quality of AI-assisted development workflows.
Provenance
The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-06-24 |
| archive | success | openalex | — | — | 5 | 2026-06-26 |
| extract | success | pdftotext | — | — | 2 | 2026-06-26 |
| clean | success | clean | — | — | 1 | 2026-06-26 |
| chunk | success | chunk | — | — | 1 | 2026-06-26 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-06-26 |
| enrich | success | openalex | — | — | 1 | 2026-06-26 |
| promote | success | — | — | — | 1 | 2026-06-24 |
| summarize | success | llm | qwen3.6-27b-prismaquant | summ-v5 | 1 | 2026-06-26 |
| tag | success | vector_similarity | — | — | 6 | 2026-06-26 |
| verify | success | — | — | — | 1 | 2026-06-26 |
Summary generated by qwen3.6-27b-prismaquant on 2026-06-26; verification: verified.
Topics
Ranked by relevance to this paper. Hover a topic for its definition.
Information type
What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).
- Methodological Resource: tool software
- Theoretical Contribution: computational model