The article discusses a significant shift in AI programming culture dubbed ‘vibe coding,’ popularized by former Tesla AI director Andrej Karpathy. This approach represents a departure from traditional software engineering practices, focusing on intuition and experimentation rather than rigid methodology. Karpathy describes it as writing code that ‘vibes correctly,’ emphasizing feel and intuition over formal computer science principles. The method has gained traction in AI development because neural networks often operate in ways that defy conventional programming logic. Key aspects include rapid prototyping, extensive experimentation, and a more fluid approach to debugging. The article highlights how this methodology has become increasingly prevalent in Silicon Valley, particularly in AI research labs and startups. Critics argue this approach might lead to less reliable or maintainable code, while supporters contend it’s necessary for the unique challenges of AI development. The piece also explores how vibe coding reflects broader changes in software development culture, particularly in the AI sector. Industry experts suggest this trend might represent a fundamental shift in how future AI systems are developed, though they caution that a balance between intuitive and traditional approaches may be optimal. The article concludes by noting that as AI continues to evolve, development methodologies will likely continue to adapt, with vibe coding potentially becoming a more formalized practice.
Source: https://www.businessinsider.com/vibe-coding-ai-silicon-valley-andrej-karpathy-2025-2