Vibe coding

hideThis article has multiple issues. Please help improve it or discuss these issues on the talk page. (Learn how and when to remove these messages) This article provides insufficient context for those unfamiliar with the subject. Please help improve the article by providing more context for the reader. (March 2025) (Learn how and when to remove this message) This article may lend undue weight to certain ideas, incidents, or controversies. Please help improve it by rewriting it in a balanced fashion that contextualizes different points of view. (March 2025) (Learn how and when to remove this message) The examples and perspective in this article may not include all significant viewpoints. Please improve the article or discuss the issue. (March 2025) (Learn how and when to remove this message) (Learn how and when to remove this message)

Vibe coding is an AI-dependent computer programming practice where a programmer describes a problem in a few sentences as a prompt to a large language model (LLM) tuned for coding. Software can be quickly created and debugged while ignoring the details of the generated code.^[1][2][3] Vibe coding is claimed by its advocates to allow even amateur programmers to produce software without the extensive training and skills which coding previously needed.^[4] The term was introduced by Andrej Karpathy in February 2025.^[5][2][4][1]

Computer scientist Andrej Karpathy, a co-founder of OpenAI and former AI leader at Tesla, introduced the term vibe coding in February 2025.^[5][2][4][1] The concept refers to a coding approach that relies on AI tools (LLMs), allowing programmers to generate working code by providing natural language descriptions rather than manually writing it.^[1] Karpathy described his approach to vibe coding as interacting with AI in a passive manner, such as through voice commands, allowing the AI to handle implementation details. He characterized the process as relying on the AI's capabilities without focusing on the underlying code.^[2] For example, he asks the AI to perform tasks such as adding padding on a sidebar. When errors arise, he copies them into the system without further explanation. He said, "It's not really coding - I just see things, say things, run things, and copy-paste things, and it mostly works."^[4] Karpathy acknowledged that vibe coding has limitations, noting that AI tools are not always able to fix bugs, requiring him to experiment with changes until the problem is resolved.^[2]

A key part of the definition of vibe coding is that the user accepts code without full understanding.^[1] AI researcher Simon Willison said: "If an LLM wrote every line of your code, but you've reviewed, tested, and understood it all, that's not vibe coding in my book—that's using an LLM as a typing assistant."^[1]

Vibe coding relies on AI tools such as Replit Agent, Cursor Composer, Pythagora, Bolt, Lovable, and Cline,^[3][6] that build on large language models from companies including OpenAI or Anthropic.^[4]

New York Times journalist Kevin Roose, who is not a professional coder, experimented with vibe coding to create several small-scale applications.^[3] He described these as "software for one", referring to personalised AI-generated tools designed to address specific individual needs, such as an app that analyzed his fridge contents to suggest items for a packed lunch. Roose noted that while vibe coding enables non-programmers to generate functional software, the results are often limited and prone to errors.^[3] In one case, the AI-generated code fabricated fake reviews for an e-commerce site.^[3] He suggested that vibe coding is better suited for hobby projects rather than essential tasks.^[3] He also observed that AI-assisted coding enables individuals to develop software that previously required an engineering team.^[3]

In February 2025, Business Insider described vibe coding as a new buzzword in Silicon Valley.^[4]

In March 2025, Y Combinator reported that 25% of startups in its Winter 2025 batch had codebases that were 95% AI-generated, reflecting a shift toward AI-assisted development.^[7][8]

Vibe coding has raised concerns about understanding and accountability. Developers may use AI-generated code without fully comprehending its functionality, potentially leading to undetected bugs and errors. While this approach may be suitable for prototyping, it is considered by some experts to pose risks in professional settings, where a deep understanding of the code is crucial for debugging and maintenance. Ars Technica cites Simon Willison, who stated: "Vibe coding your way to a production codebase is clearly risky. Most of the work we do as software engineers involves evolving existing systems, where the quality and understandability of the underlying code is crucial."^[1]

In March 2025, Forbes reported that while vibe coding has accelerated software development, it presents challenges for long-term software reliability. The article notes that AI-generated code can be difficult to troubleshoot, often requiring human oversight. Forbes states, “Companies embracing vibe coding must develop robust code review processes and invest in debugging tools to ensure quality and reliability.” Additionally, the article questions how this approach will impact software security and maintainability over time.^[8]

• Vibe Coding subreddit