The guy who coined vibe coding says AI is making programming 'unrecognizable'

Andrej Karpathy, a notable figure in the AI landscape, is making waves with his insights into the rapid evolution of programming. Known for coining the term "vibe coding," Karpathy has an impressive resume that includes being a founding member of OpenAI, leading AI initiatives at Tesla, and eventually returning to OpenAI to establish his own lab. In a recent post on X, Karpathy highlighted a remarkable acceleration in what he refers to as "agentic coding," which has dramatically changed the programming scene in mere months. "Coding agents basically didn't work before December and basically work since," he stated, emphasizing the disruptive nature of these new tools on traditional programming workflows. Karpathy shared a personal example to illustrate this shift. Over a weekend, he utilized AI to create a comprehensive video analysis dashboard for his home security cameras. Remarkably, his AI agent completed the task within 30 minutes, autonomously troubleshooting errors along the way. "Programming is becoming unrecognizable," he declared, suggesting that the industry is far from the status quo. He has become known for his extensive, thoughtful commentary on software engineering. In a recent discussion, he addressed concerns about whether large teams of engineers would be supplanted by a handful of skilled prompters. Responding to an AI developer's query, he noted that while vibe coders are making strides, deep technical knowledge may become even more valuable due to the enhanced leverage AI provides. Karpathy's terminology has sparked debate among industry peers. Notably, Peter Steinberger, creator of OpenClaw, criticized the term "vibe coding," suggesting it diminishes the complexity and effort involved in coding with AI. Karpathy countered this sentiment by referencing a mantra from his time at Tesla: "Every action is error," underlining the importance of structuring AI agents to optimize workflow and eliminate bottlenecks. As the conversation around AI in programming continues to evolve, Karpathy reminds us that while these tools may seem like magic, they fundamentally rely on high-level guidance and a discerning sense of taste, framing the future of coding as a strategic act of delegation rather than an effortless task.