In the programming world, there’s a rising trend some call vibe coding — a term popularized by Andrej Karpathy, meaning you let AI agents write all the code for you, blindly, without personally reviewing or understanding it. You just “vibe,” prompt after prompt, never stopping to check, learn, or correct the output yourself.
Let me be clear: I’m not talking about using LLMs, RAG, or prompting to assist coding, generate snippets, or look up APIs. That’s not new. We’ve always worked like this — from reading manuals, to hunting Stack Overflow in the 2010s, to asking LLMs today. What’s changed is the speed and specificity of the help, not the core idea.
The real productivity gain with AI is that we can offload tedious typing and spend more time doing things we once skipped: writing more tests, reviewing code carefully, improving maintainability, adding comments, and making sure the codebase is clean and efficient.
In my experience, no software survives untouched after its first draft. There’s always another feature, another change, another revisit. If you build something you don’t understand, you will struggle to extend or fix it later.
For anyone learning to code today, I strongly recommend this:
- Use AI to assist, but don’t surrender.
- Learn what the AI outputs mean.
- Write some code without AI, too.
Programming isn’t about memorizing every syntax detail; it’s about building a mental model of how things work. Anyone can learn to code — no gatekeeping here — but you owe it to yourself (and your future projects) to understand what you’re shipping.
So, no, “vibe coding” doesn’t work for me. And if you want to truly grow as a developer, I believe it won’t work for you either.