Jomkit Jujaroen

An Argument for Vibe Coding

AI facing off against a software engineer
Back

Written by: Jomkit Jujaroen

First Published: 15/08/2025

Image is AI-generated

Intro

To vibe code—or not to vibe code? In other words, should we let AI generate our code while we guide the process, or should we stick to typing it all ourselves?

When LLM-powered IDEs first arrived, I was skeptical—just like many other developers. Could an AI really write usable code? Would it make us lazy? Or worse, replace us entirely? Early adopters, however, saw the writing on the wall: vibe coding isn’t just a trend. It’s here to stay.

Will LLMs Replace Software Engineers?

That’s like asking if calculators made mathematicians obsolete. Of course not. Calculators didn’t erase math—they freed mathematicians to focus on deeper, more complex problems.

LLMs and the Role of Critical Thinking

LLMs are much the same. They remove tedious boilerplate, but without the knowledge to know which boilerplate is needed and why, they’re useless. The value lies in the critical thinking you bring to the table.

The Difference Between Juniors, Seniors, and LLMs

The gap between a junior and senior engineer isn’t raw speed or syntax knowledge—it’s wisdom.

AI Is a Tool—Not a Replacement

Anyone who’s used AI assistants knows their flaws: hallucinations, excessive compliance, and the occasional “Yes, that’s correct” when it’s not. That’s why your role as a reviewer and skeptic is irreplaceable.

Some developers dismiss vibe coding because it still requires supervision. But that’s like refusing to use a calculator because you can multiply 412.32 × 43.12 by hand. Sure, it’s possible—but it’s slower. Whether LLMs make you 2× or 10× faster, faster is still faster.

The Real Danger

The real danger isn’t in using AI—it’s in forgetting the fundamentals. If you rely on it so much that you lose your ability to debug, architect, and think critically, then you’re in trouble.

Vibe coding won’t replace you. But ignoring it might leave you behind. Use it to work faster, learn faster, and focus on the parts of engineering that only a human mind can do.