AI & Tech 29

Why are people against it? Sounds like it'll only grow the tech pool.

The pushback isn't really against vibe coding growing the tech pool, which it genuinely does: the concern is that code built without understanding can be quietly dangerous the moment real users touch it.

Answer from Mary's digital twin. Checkmark turns Green once she confirms it.

The critics aren't professional gatekeepers who want to keep outsiders out. Most of them have shipped something in production and watched it break in an unexpected way. Their worry sits in a few concrete places.

Security you can't see

AI-generated code can appear functionally correct while hiding critical flaws. A generated SQL query might return the right data in testing but fail to prevent injection attacks. An API integration might work in a demo but expose credentials in logs. The code runs, so it looks right. The vulnerabilities aren't visible until something breaks. Since many of the people using vibe coding are not experienced developers, there's a real chance they'll miss security problems before going live. The gap between "it works on my machine" and "it is safe for 10,000 people's data" is exactly where things go wrong.

Code nobody understands, including you

A related concern is the potential loss of codebase understanding at the team level. When developers rely heavily on AI-generated code without deeply reviewing it, institutional knowledge of how a system works can erode. The less a team understands its own codebase, the harder it becomes to debug, extend, or secure it over time. Codebases like these accrue technical debt quickly. AI can introduce inefficient or overly complex solutions without considering long-term design, and because the AI doesn't have a thorough understanding of system architecture, it can produce inconsistent naming, coding styles, or logic flows. Technical debt, by the way, is the hidden cost of cutting corners now: every shortcut that works today becomes a harder fix tomorrow.

The junior-developer trap

Some worry that over-reliance on vibe coding could cause programmers, especially junior ones, to insufficiently learn key programming concepts, a risk that can facilitate knowledge-related mistakes in vibe-coded projects. There's a real difference between being able to prompt your way to working code and understanding why it works. If the AI hits a wall, someone has to know what to do next.

What the critics are actually saying

Supporters treat vibe coding as a powerful, quick way to develop prototypes, while experienced engineers see it as an anti-pattern that trades quality, security, and maintainability for speed. A large middle group considers using it with caution: fine for experiments or hobbyist projects, but too risky for production.

That middle-ground is the honest answer. Vibe coding to build a personal tool, test a business idea, or get a prototype in front of investors is genuinely powerful, and it does widen who gets to participate. Shipping that same prototype to real users, with real data, without a review pass, is where the friction is legitimate. The tool isn't the problem. Treating a prototype as production-ready is.

Comments

Comments are public. Links are removed.

Follow-ups

Ask a follow-up
Answered instantly. We don't show your name.
This follows on from
What's vybe coding?

Vibe coding is building real software by describing what you want in plain English, and letting an AI write all the actual code for you.

Read that one first

Build something real, starting from where you are.

CEEK is a community of creators who are figuring out the tools, the money, and the next move, with daily posts from people doing it right now.

Join me on CEEK