Vibe Coding Still Needs Someone Who Can Code

AI can write code now. You describe what you want, and a tool builds it for you, with no deep programming knowledge required. People call this vibe coding, and it has caught on fast, including in how websites get built. At its recent Search Central Live event in Milan, Google addressed it head on. The message was less a celebration than a caution.

Building a site by describing it

Vibe coding is what it sounds like. Instead of writing the code yourself, you tell an AI tool what you want and let it generate the result. Need a contact form, a landing page, a small script to pull some data? Describe it in plain language and the tool writes it. For people who can’t code, this feels like magic. For people who can, it’s a fast way to skip the boring parts.

It works well enough that a lot of sites now get built this way, often on JavaScript frameworks the AI reaches for by default. Google zeroed in on that. The code runs, the site loads, and everything looks fine on the surface. What’s underneath is where the trouble can hide.

Google’s actual message

Google put up a slide titled vibe coding plus SEO, and the points on it were blunt. You should still know how to program. Vibe coding can replace some tools, but watch the long-term time and security cost. Use official APIs like the Search Console API instead of hacking together your own scraper. Plenty of vibe-coded sites run on JavaScript frameworks, so know what to look out for there. And the slide tied it together with one line. Technical SEO skills are more important than ever.

None of that reads like a company telling people to hand the work to AI and walk away. Google builds AI tools, and it still said, in effect, learn the craft anyway. The tools work as an accelerator for people who already know what they’re doing, rather than a replacement for the knowledge.

The costs that show up later

The warning about cost is the heart of it. AI can produce working code in seconds, but working is not the same as solid. A vibe-coded site can carry security holes the person who built it never sees, because they never read the code closely enough to spot them. Maintenance gets harder too. When something breaks, fixing code you don’t understand is slow and risky. And the architecture, the way the whole thing fits together, tends to get messier the more you let a tool make those calls.

The JavaScript framework point is a specific version of this. AI tools love to reach for heavy frameworks, and those can cause real SEO problems when nobody is watching. Content that leans on JavaScript to load can render slowly, resist crawling, or go missing entirely if it’s built wrong. A product page where the descriptions only show up after a script runs, for instance, can look complete to a visitor and come back blank to a crawler. Someone who knows technical SEO catches that. Someone who shipped what the AI handed them often doesn’t, until traffic drops and they go looking for why.

The point about official APIs lands the same way. When you need data out of a tool like Search Console, the supported API is built for it and stays stable. A scraper an AI tool throws together to grab the same numbers might work today and break next week, hit rate limits, or run afoul of the terms of service. Reaching for the proper endpoint is slower to set up and a lot less painful later.

The skills become the real advantage

For anyone working in SEO, there’s an upside in all of this. As AI does more of the building, the people who understand what’s being built get more valuable, not less. When anyone can generate a site in an afternoon, the edge goes to whoever can tell a good build from a broken one, spot the security risk, fix the framework issue, and use the right tools the right way.

Google added a related warning. Be critical of vibe-optimization for AI. The same trap shows up in content, not only in code. In the same way you can vibe-code a site without understanding it, you can vibe-optimize content by chasing surface-level tweaks that sound clever and do nothing. We’ve written before about the hacks Google says to ignore, and this fits right in. Surface-level moves don’t earn visibility, but substance does.

For anyone deciding how far to lean on AI tools, Google’s framing is a healthy reality check. The tools are real and useful. They also reward the people who know enough to use them well and to catch what they get wrong.

The same logic runs through how a brand earns visibility in search and in AI results. No tool shortcuts authority. Link building and digital PR build the kind of trust signals an AI tool can’t fake on your behalf, and they take real work and judgment. Vibe coding can help you ship faster, but it can’t decide whether what you shipped is any good. That part still belongs to people who know what they’re doing.