Vibe coding debate heats up as AI reshapes how software gets built

Summary

A fierce debate over so‑called “vibe coding” is splitting the software world as AI tools move to the center of how code is written. Some see it as a productivity boost and a way to open software creation to more people, while others warn about security, quality and long‑term damage to the profession.

Emergent, a startup built around this style of AI‑driven development, has become a showcase for the optimistic view after reaching strong recurring revenue in just a few months. At the same time, academics and career experts are sounding alarms about what this shift might do to open source communities and to how new developers learn their craft.

In practice

“Vibe coding” describes a way of building software where people describe what they want in natural language and let AI systems generate most of the code. Instead of writing each line, the developer nudges the result with iterative prompts, often without closely tracking the low‑level details of the implementation.

On the business side, Emergent illustrates the pull of this model: the company hit 100 million dollars in annual recurring revenue in eight months, backed by major investors and millions of users with no prior coding experience. For many of those users, turning an idea into a working app through a handful of instructions feels like a new on‑ramp into software.

The risks are already visible. Security researchers have documented dozens of vulnerabilities in applications created with AI coding agents, including weak authentication and missing input validation. Studies find that while a solid share of AI‑generated solutions are functionally correct, only a small fraction meet basic security standards.

There are also signs of strain in the wider software ecosystem. One academic paper argues that vibe coding could “kill” open source by cutting direct engagement with maintainers and technical forums, pointing to evidence like the drop in questions on Stack Overflow after large language models arrived. For junior developers, critics warn that leaning too heavily on AI may stunt the hard‑won skills of debugging, architecture and understanding how systems work end‑to‑end.

Not everyone sees the trend as harmful. Some programmers argue that if teams review, test and genuinely understand AI‑generated code, vibe coding can function as a kind of “typing assistant” rather than a crutch. Leaders at AI companies also say they already rely on agents such as Claude for the majority of their internal code, while human engineers stay in charge of system design and the hardest decisions.

Who benefits / who loses

In the short term, vibe‑coding platforms and the organizations that adopt them aggressively stand to gain speed and broaden who can contribute to software projects. Startups and small businesses can spin up prototypes and internal tools without large engineering teams, shifting the balance of power in software production.

On the other side, open source communities, early‑career developers and security teams may lose out if the trend advances without guardrails. Less hands‑on contact with code and with technical communities could slow knowledge transfer, deepen the maintainer crunch and increase the risk of buggy and vulnerable systems in the wild.

Why it matters

• Vibe coding highlights how AI is shifting work from manually writing code to explaining and reviewing solutions.
• The fast growth of dedicated platforms signals strong demand for tools that democratize app building.
• Documented security issues and pressure on open source raise questions about how to adopt these tools responsibly.
• The skills debate around junior developers could reshape how we train and hire in software over the next decade.