Google DeepMind has connected its general-purpose world model, Project Genie, with Google Street View, allowing users to generate interactive simulations of real-world locations. The new capability, announced at the Google I/O 2026 developer conference, moves beyond static 360-degree photos to create dynamic environments that can be explored and modified.
Quick Facts
- Genie world model now uses Street View data.
- Creates interactive simulations of real locations.
- Announced at Google I/O 2026 developer conference.
From Static Images to Interactive Worlds
For years, Google has amassed a colossal visual library of the planet, with its Street View cars and “tracker backpacks” capturing over 280 billion images across 110 countries. Now, instead of just passively viewing these locations, Genie can use that data as a foundation to build explorable virtual sandboxes.
“With Street View, we have imagery from a large quantity of the world,” Jack Parker-Holder, a research scientist on DeepMind’s open-endedness team, told TechCrunch. “You can imagine how potentially powerful it is to combine this rich source of real-world information and data with an ability to simulate worlds.”
The integration allows users to generate a simulation of a real block and then alter variables, like changing the weather to see what a familiar street looks like covered in snow. Jonathan Herbert, director of Google Maps, noted that while Genie can’t yet create a perfect reconstruction, its key breakthrough is spatial continuity. The model correctly remembers and simulates the environment in a full 360-degree turn, providing a stable base to build upon.
Training Robots and Autonomous Systems
One of the most powerful applications for this technology is in training AI agents and robotics. Waymo, Google’s self-driving car unit, is already using Genie 3 to train its AI driver on “exceedingly rare events” like tornadoes or unexpected animal encounters. Adding real-world Street View data can help Waymo prepare its systems for launches in new cities globally.
Parker-Holder offered another example: training a new robot for deployment in London. “Genie could,” he explained, “simulate those scarce occasions when the sun glints off the Victorian housing, so the rays don’t shock the robot when it happens.” This ability to simulate specific, rare environmental conditions in real-world locations is a significant step forward for training robust AI systems.
What This Means for MENA’s Tech Scene
For the MENA region, the implications of Genie’s new capability are substantial. As nations like the UAE and Saudi Arabia invest heavily in smart cities and autonomous transport for projects like NEOM, the ability to train AI drivers and delivery robots in hyper-realistic simulations of local environments—complete with unique architectural styles and weather conditions like sandstorms—is a critical advantage.
Beyond smart cities, the region’s rapidly growing gaming and entertainment sector stands to benefit. Local developers could use Genie to create immersive games set in authentic, explorable versions of cities like Dubai, Cairo, or Riyadh, lowering development barriers for creating realistic open-world environments. The technology also presents opportunities for PropTech and tourism, allowing developers to visualize new projects in existing cityscapes or create interactive previews of historical landmarks.
A Glimpse of the Future, But Still an Experiment
Despite the impressive progress, Google’s team emphasizes that the technology is still experimental. Diego Rivas, a product manager at DeepMind, cautioned that there is much to improve in terms of accuracy. The current simulations are recognizable but have a video-game quality rather than being photorealistic.
Crucially, the models are not yet physics-aware. A demo showed a character running through cacti and bushes in a simulated Joshua Tree, highlighting that the AI doesn’t yet understand cause and effect. Parker-Holder estimates the model is “six to 12 months behind video in terms of the accuracy and quality.”
Google is rolling out Street View in Genie to select Google AI Ultra subscribers in the United States, with global access planned over the next few weeks.
About Google DeepMind
Google DeepMind is an AI research laboratory and a subsidiary of Google. Its mission is to “solve intelligence” by building advanced artificial intelligence systems. The lab focuses on fundamental AI research to advance science and benefit humanity, with breakthroughs in areas ranging from gaming and protein folding to large-scale language models.
Source: TechCrunch
