At a recent firechat session during the LangChain Interrupt conference in May, Stanford professor and former Google Brain scientist Andrew Ng expressed his concerns about the term “vibe coding.” The practice, which has become a buzzword for coding with AI assistance, has been described by some as a relaxed and intuitive approach to software development. However, Ng argued that the name is misleading, as it suggests that developers are simply “going with the vibes” when using AI tools. “It’s unfortunate that that’s called vibe coding,” he said. “It’s misleading a lot of people into thinking, just go with the vibes, you know — accept this, reject that.”
Ng emphasized that coding with AI is actually a deeply intellectual exercise that requires significant mental effort. “When I’m coding for a day with AI coding assistance, I’m frankly exhausted by the end of the day,” he said. While he is critical of the term, Ng remains positively disposed towards the technology itself, acknowledging its potential to significantly enhance software development. “It’s fantastic that developers can now write software faster with these tools,” he stated. “Sometimes while barely looking at the, code” he added, indicating that AI-assisted coding can boost productivity without requiring constant oversight.
Ng’s comments have sparked a discussion within the tech community about the terminology used to describe AI-assisted coding practices. Some developers have embraced the term “vibe coding” as a way to highlight the efficiency and ease of AI-assisted development. However, Ng’s critique highlights the need for more accurate language that reflects the complexity and demands of the work. He suggested that the field should move towards more precise terminology that better represents the intellectual rigor required when using AI coding tools.