Vibe Coding: Love It or Hate It

All I wanted was to build my own website. Vibe coding felt like the perfect way to do it: no “middlemen” and no templates. The freedom to change my site every day if I wanted to felt unreal. I started learning about Cursor, watching tutorials and slowly figuring things out.
So I vibe-coded this page.
“Don’t code by hand. Get AI to help you code, and that will make people in all job functions much more productive and have more fun.
People that code, be it CEOs and marketers, recruiters, not just software engineers, will really get more done than ones that don’t”
Andrew Ng, founder of Google Brain
The start
I opened Cursor, outlined the structure I had in my mind and the AI generated an initial version of the site within minutes. It was not a completed product (of course) but a starting point and now the more detailed work began. Even so, having a functional framework appear so quickly created an immediate sense of flexibility and activated this “desire to build” function known the feeling that you can shape and create something.
From there, I could change things simply by describing the adjustments I wanted:
“Refactor this layout into a two-column grid”
“Optimize the header spacing for better visual hierarchy”
…and the changes appeared instantly.
For the first time I was not constrained by a template or reliant on a developer to interpret my ideas. If I wished to adjust typography late in the evening or rework an entire page the next morning, I could do it freely.
Security & trust
As I learned from the videos - vibe coding without safeguards is risky. Andrew Ng’s message is very optimistic but if you plan to bring your project online you also need to review its security.
AI-generated code can carry unseen vulnerabilities. For example if you say: “Create a login form,” you might get code that stores passwords in plaintext or leaves API keys exposed.
So here’s what I learned:
Of course I’ve also come across many sceptical articles about vibe coding but that is precisely why it’s worth learning. New tools always bring both possibilities and risks and the only way to understand the difference is to develop the competence to use them.