Google DeepMind CEO Demis Hassabis Hints At Simulation Theory After New Product Release

As AI is getting stronger, simulation theory seems to be gaining more prominence in everyday conversation.

Google DeepMind CEO and 2024 Nobel Prize winner Demis Hassabis has hinted at the possibility of simultation theory after a new product release from Google. “Nature to simulation at the press of a button, does make you wonder…,” he posted on X, responding to a Google Gemini video showing how the latest AI model could create computer code to represent everyday objects, like trees and spider webs.

“We’re releasing an updated Gemini 2.5 Pro (I/O edition) to make it even better at coding,” Google DeepMind had earlier posted. “You can build richer web apps, games, simulations and more – all with one prompt. In GeminiApp, here’s how it transformed images of nature into code to represent unique patterns,” it added. The accompanying video showed how users could upload a picture of a tree with a prompt “transform this image into a code-based representation of its natural behaviour”. Gemini not only understood that the image was a tree, but coded one from the ground up, starting from the roots, then the trunk, and finally branches and leaves. Similarly, Google was able to write code to create spiderwebs as they were being built, and also simulate natural processes like flames in a fire.

This appeared to make Hassabis wonder if natural processes that we see around us might themselves be running through some sort of code. This is the basis of simulation theory, which says we’re currently living in a computer simulation, much like characters in a video game. The theory says that once civilizations become sufficiently advanced, they’re able to create realistic simulations of artificial worlds, like we’ve done with immersive computer games. Given that the universe is really old, not only would’ve many civilizations created simulations, but these simulations would’ve created simulations of their own, and those simulations would’ve created even more simulations. This means that there are likely millions of such simulations, and just one base reality. On the pure probability basis, we’re much more likely to be living in one of the simulations than the singular base reality.

It’s an idea that’s been previous put forward by Elon Musk, who’s often said that there’s a good chance we’re all in a simulation. Epic Games CEO Tim Sweeney has said that as AI becomes more powerful, it makes ideas like simulation theory harder to dismiss. And if we can create fake worlds, like Gemini can code trees and spiderwebs instantly through code, who’s to say that we aren’t living in a similarly simulated world ourselves.