Gemini 3 and the Rise of Generative UI: A Revolution or Chaos for the Web?

Google introduced last week Gemini 3 and among all the innovations, one was lost that might be more important than improved language skills. It’s about being able to create generative UI and Google’s vision is to change the way we use apps and websites. Instead of static interfaces that are pre-programmed, you get dynamically created environment exactly for your query.

It sounds like science fiction, but Google is already deploying it in gemini. And the debate between developers and users is surprisingly heated – some see a revolution, others a disaster.

What is generative UI and how does it work?

Imagine you are looking for information about a mortgage.Today you visit the bank’s website, scroll through a static page, find a calculator, do some calculations, then open a table in Google Sheets you switch to notes, maybe even to the calendar. Generative UI all this combine into one dynamically created interface.

In Gemini 3 Pro you can choose between Dynamic view a Visual view. The first variant uses the ability of AI to write code and create your own user interface – an interactive calculator, a physics simulation, or even a mini-request. Visual view will then offer sliders and filters to customize the content.

For now, it works so you have to turn the mode on manually, but Google is already hinting at a future where AI will recognize when gen UI is a better choice than a classic text response.

Reddit has a different take: the end of static sites or the beginning of chaos?

an interesting debate has developed on Reddit, started by a developer with the thesis that generative UI is killing static websites faster than we think. He gives an example: you are looking for a plumber at 2 in the morning. Instead of a generic home page with a menu, you get a dynamically created page with with the giant red button “Call dispatch 24/7” and no extra navigation.

But the reactions are very skeptical. One of the top rated comments says: “Websites are more than raw information. They’re experiences.” They contain pictures, videos, information that you were not even looking for. It’s more complicated than Google makes it out to be.

Another user points out a practical problem: “I wouldn’t trust an AI button that says ‘call a plumber’. What if it connects me to a pizza place?” And he’s right – AI hallucinations are still a problem and can be risky in critical services.

Technical challenges

The develo