Vibecoding, AI, and why programming still matters

Lately, I’ve been seeing a wave of posts about “vibecoding” and how AI is apparently going to replace developers. There’s this idea floating around that young people shouldn’t even bother learning to code because “developers are doomed.” Meanwhile, folks with zero dev experience are sharing videos of themselves spinning up full-stack apps using nothing but prompts. It’s everywhere.

But let me be clear—things are not that simple.

Wikipedia says that Vibecoding is an AI-dependent technique where a person describes a problem in a few sentences as a prompt to a large language model (LLM) tuned for coding. Vibe coding is claimed by its advocates to allow everyone to produce software without the extensive training and skills required for software engineering.

The reality is that programming without understanding systems, architecture, or fundamentals will only get you so far. Large Language Models are incredible tools—and they’re only getting better—but experience, skill, and actual problem-solving still matter. A lot.

Over the past few weeks, I tried vibecoding trend myself. I tried generating UIs with React, building out backends with Node.js, even messing around with smart contracts. All using prompts. And honestly? It wasn’t that exciting.

Yes, AI can handle boilerplate. Yes, it can spit out components or even entire apps. But in my opinion? The quality of that code was garbage. We’re talking PRs with 99+ comments. Fragile code. Weird patterns. Stuff that even junior devs would raise an eyebrow at. And don’t even get me started on bug fixing—bouncing between three AI-generated bugs, all of which I could’ve fixed faster manually.

But here’s the twist.

What did impress me was using AI inside a codebase I already know and care about. A codebase I’ve nurtured, where I understand every line. In that environment, AI is a game-changer. It boosts dev velocity in a very real way. It helps navigate, suggest, accelerate—but doesn’t replace the thinking behind the build.

So here’s what I believe:

Young people should learn programming. In fact, there’s never been a better time. With AI tools, you now have a superpower for learning—instant explanations, examples, and debugging help.

And for experienced developers? This is not the time to slow down or worry about being replaced. Now’s the time to double down. Invest in your skills. Learn how to leverage AI, not be replaced by it. Use it to boost your speed, but don’t compromise on code quality or system thinking.

AI is a powerful assistant—but real engineering still matters.