Skip to content

Pick your style: Denkbaar

Book a conversation

Articles

History

Genie 3 shows worlds that respond to the user

·

Google DeepMind presented a model for generated environments that users can control in real time.

A worker controls a walk through a drawn room on a screen.
A generated environment can respond during use. Illustration: Denkbaar.

On 5 August 2025, Google DeepMind presented Genie 3. The company described it as a world model: a model that can generate an environment in which a user can move. The announcement identifies real-time interactive worlds as the core of the news. The output was therefore not a fixed image or a predetermined film, but an environment that responds during use. Source. Two screens show the same room from different positions.

This made the announcement interesting as a step in creating interactive media. A description became the starting point for an experience. For a business, it raises a different design question: what should someone be able to do to understand an idea? Two designers compare a floor plan with a walkthrough on a screen.

For a small or medium-sized business, the lesson may be to choose the intended action before building a demonstration. As a design exercise, consider a visitor walking through a proposed room. Decide which parts must be fixed and which may only suggest an atmosphere. Do not treat this historical demonstration as evidence that dimensions or a real working situation will be reproduced accurately. For a concrete application, request a trial with clear boundaries. Then judge whether the interaction explains something that an ordinary image cannot make clear.

Sources

  1. Google DeepMind Genie 3 world model

Topics: modelhistory