Silicon Valley is embracing a revolutionary approach to software development called “vibe coding,” a term coined by Andrej Karpathy, OpenAI cofounder and former Tesla AI leader. This method represents a fundamental shift in how code is created, with developers using AI to write software through simple prompts and conversations rather than traditional line-by-line programming.
Karpathy describes vibe coding as an approach where developers “fully give in to the vibes” and “forget the code even exists.” In a February 2025 post on X, he explained: “It’s not really coding — I just see stuff, say stuff, run stuff, and copy-paste stuff, and it mostly works.” This defies conventional wisdom that software development requires virtuosic engineering skills.
The practice leverages advanced AI tools like OpenAI’s Composer (developed with Cursor AI) and Anthropic’s Sonnet model. Karpathy demonstrates using these tools alongside voice-to-text AI like Superwhisper, allowing him to create applications without “barely even touch[ing] the keyboard.” When errors occur, he simply copy-pastes error messages back to the AI, which typically fixes them automatically.
Industry leaders are predicting major disruptions to traditional software engineering roles. OpenAI CEO Sam Altman stated during a February 2025 India visit that software engineering would be “very different by the end of 2025.” Meta CEO Mark Zuckerberg told Joe Rogan that AI would soon handle the work of midlevel engineers at his company.
The trend is already showing significant adoption. Amjad Masad, CEO of AI-backed software company Replit, revealed that “75% of Replit customers never write a single line of code.” Startups like Menlo Park Lab are fully embracing the methodology, with founder Misbah Syed using vibe coding to build products like Brainy Docs, which converts PDFs to explainer videos.
However, experts warn of potential downsides. Harry Law, an AI researcher at Cambridge University, notes that “ease of use is a double-edged sword” that might prevent beginners from learning system architecture or performance optimization. Security vulnerabilities and technical debt are additional concerns. A senior Microsoft engineer called the concept “a little overhyped,” noting that large language models “get lost in requirements and generate a lot of nonsense content” for complex projects. Venture capitalist Andrew Chen from Andreessen Horowitz described the experience as “both brilliant, and enormously frustrating,” noting that while the first 75% comes easily, iterating and making changes becomes challenging.
Key Quotes
It’s not really coding — I just see stuff, say stuff, run stuff, and copy-paste stuff, and it mostly works.
Andrej Karpathy, OpenAI cofounder and former Tesla AI leader, describing his vibe coding approach on X. This quote encapsulates the radical simplification of software development that AI tools are enabling, challenging traditional notions of what programming requires.
The hottest new programming language is English.
Karpathy said this less than two months after ChatGPT’s release in late 2022, presciently predicting how natural language prompting would transform coding. This observation has proven increasingly accurate as AI coding assistants have advanced.
75% of Replit customers never write a single line of code.
Amjad Masad, CEO of Replit (backed by A16z and Y Combinator), revealed this striking statistic in response to Karpathy’s post. It demonstrates how widespread AI-assisted development has already become and validates that vibe coding is not just theoretical but actively practiced at scale.
For a total beginner who’s just getting a feel for how coding works, it can be incredibly satisfying to build something that works in the space of an hour.
Harry Law, AI researcher at Cambridge University, acknowledging the democratizing potential of vibe coding while also warning that this ease of use is “a double-edged sword” that may prevent developers from learning fundamental concepts like system architecture.
Our Take
Vibe coding represents both the promise and peril of AI’s integration into professional workflows. While Karpathy’s enthusiasm is compelling, the skepticism from working engineers reveals a critical gap between prototype-level development and production-grade software. The 75% statistic from Replit is impressive but raises questions about code quality, maintainability, and scalability.
The real transformation may not be replacing engineers but bifurcating the profession into those who can architect complex systems and guide AI tools effectively versus those who simply execute predefined patterns. Andrew Chen’s observation about getting “the first 75% trivially” but struggling with iteration perfectly captures this limitation. As AI coding tools mature, the industry will need to develop new best practices, review processes, and educational frameworks that balance accessibility with engineering rigor. The vibe may be changing, but fundamental software principles remain essential.
Why This Matters
Vibe coding represents a potential paradigm shift in software development that could democratize programming while disrupting traditional engineering careers. The ability for non-programmers to build functional applications through conversational AI fundamentally challenges the barriers to entry that have historically protected software engineering as a specialized profession.
This trend has profound implications for the tech workforce. With industry leaders like Sam Altman and Mark Zuckerberg predicting dramatic changes by late 2025, companies may restructure their engineering teams, potentially reducing demand for mid-level developers while increasing need for AI prompt engineers and senior architects who can guide AI systems effectively.
The technology also signals AI’s maturation from assistant to autonomous creator. The fact that 75% of Replit customers never write code demonstrates that AI coding tools have crossed a usability threshold where they’re genuinely accessible to non-technical users. However, concerns about technical debt, security vulnerabilities, and maintainability suggest the industry must balance innovation with engineering rigor. This tension will likely define the next phase of AI-assisted development.
Recommended Reading
For those interested in learning more about artificial intelligence, machine learning, and effective AI communication, here are some excellent resources:
Recommended Reading
Related Stories
- The Impact of AI on Software Engineering Jobs and Market Outlook
- OpenAI CEO Sam Altman’s Predictions on How AI Could Change the World by 2025
- OpenAI CEO Sam Altman Hints at Potential Restructuring in 2024
- Tech Workers Are the Real Winners in the AI Talent War, With Pay Set to Soar by 2024
Source: https://www.businessinsider.com/vibe-coding-ai-silicon-valley-andrej-karpathy-2025-2