Detailed Analysis
Boris Cherny, the creator of Claude Code at Anthropic, has publicly expressed frustration with the term "vibe coding," a phrase that has become ubiquitous in discussions about AI-assisted software development. The term, coined by prominent AI researcher Andrej Karpathy in early 2025, describes a mode of programming in which developers lean almost entirely on AI models to generate code while providing only loose, intuitive direction — essentially surrendering precise technical control in favor of iterative, conversational prompting. Cherny's objection signals a meaningful tension within the AI coding community itself: between those who frame AI tools as liberation from technical rigor and those who see that framing as a misrepresentation of what serious AI-assisted development actually involves.
Cherny's position is particularly notable given that Claude Code, the product he created, sits at the center of this debate. Claude Code is Anthropic's terminal-native agentic coding tool, designed to perform complex, multi-step software engineering tasks — writing, editing, debugging, and executing code across large codebases with meaningful autonomy. That its own creator would push back against "vibe coding" as a descriptor suggests he views the phrase as trivializing or even misleading about the depth of engineering judgment still required from developers who use tools like Claude Code effectively. The distinction he appears to be drawing is between casual, low-accountability AI prompting and disciplined, high-context collaboration with AI systems.
The broader context here is a rapidly evolving debate about what software engineering looks like in an era of capable AI agents. "Vibe coding" became popular partly because it captured something real — that AI tools have lowered the barrier to producing functional code dramatically — but critics argue the term encourages a passive relationship with code that can lead to insecure, unmaintainable, or simply incorrect software at scale. Cherny's objection fits into a growing counternarrative from professional engineers who argue that AI amplifies engineering skill rather than replacing the need for it, and that the most valuable use of tools like Claude Code still demands deep technical understanding to verify outputs and guide the model effectively.
Anthropic's positioning of Claude Code as a serious engineering tool rather than a shortcut reflects the company's broader strategy of targeting professional developers and enterprise customers who need reliable, auditable outputs. The company has invested heavily in Claude Code's ability to handle long-context, multi-file codebases and to operate autonomously within developer workflows — capabilities that go well beyond the playful spontaneity implied by "vibe coding." Cherny's public distaste for the phrase is thus also implicitly a product positioning statement, differentiating Claude Code from lighter-weight, consumer-facing AI coding assistants that may more readily earn that label.
The dispute over terminology ultimately reflects a deeper question about professional identity and accountability in software development. As AI coding tools become more capable, the industry is actively negotiating what it means to be a programmer, what skills remain essential, and how credit and responsibility for software quality should be assigned. Cherny's stance — coming from someone who has built one of the most powerful AI coding agents available — carries weight in that conversation, suggesting that even the people building these transformative tools believe the human engineer's role remains substantive, rigorous, and decidedly not something that should be reduced to a vibe.
Read original article →