Google DeepMind has revealed Genie 3, its newest basis world mannequin that the AI lab says presents an important stepping stone on the trail to synthetic normal intelligence, or human-like intelligence.
“Genie 3 is the primary real-time interactive normal function world mannequin,” Shlomi Fruchter, a analysis director at DeepMind, mentioned throughout a press briefing. “It goes past slim world fashions that existed earlier than. It’s not particular to any specific setting. It could generate each photo-realistic and imaginary worlds, and every thing in between.”
Genie 3, which continues to be in analysis preview and never publicly out there, builds on each its predecessor Genie 2 – which might generate new environments for brokers – and DeepMind’s newest video era mannequin Veo 3 – which reveals a deep understanding of physics.
With a easy textual content immediate, Genie 3 can generate a number of minutes – up from 10 to twenty seconds in Genie 2 – of numerous, interactive, 3D environments at 24 frames per second with a decision of 720p. The mannequin additionally options “promptable world occasions,” or the power to make use of a immediate to vary the generated world.
Maybe most significantly, Genie 3’s simulations keep bodily constant over time as a result of the mannequin is ready to bear in mind what it had beforehand generated – an emergent functionality that DeepMind researchers didn’t explicitly program into the mannequin.
Fruchter mentioned that whereas Genie 3 clearly has implications for instructional experiences and new generative media like gaming or prototyping inventive ideas, its actual unlock will manifest in coaching brokers for normal function duties, which he mentioned is important to reaching AGI.
“We expect world fashions are key on the trail to AGI, particularly for embodied brokers, the place simulating actual world eventualities is especially difficult,”Jack Parker-Holder, a analysis scientist on DeepMind’s open-endedness crew, mentioned throughout a briefing.
Techcrunch occasion
San Francisco
|
October 27-29, 2025
Genie 3 is designed to resolve that bottleneck. Like Veo, it doesn’t depend on a hard-coded physics engine. As a substitute, it teaches itself how the world works – how objects transfer, fall, and work together – by remembering what it has generated and reasoning over very long time horizons.
“The mannequin is auto-regressive, that means it generates one body at a time,” Fruchter informed TechCrunch in a separate interview. “It has to look again at what was generated earlier than to resolve what’s going to occur subsequent. That’s a key a part of the structure.”
That reminiscence creates consistency in its simulated worlds, and that consistency permits it to develop a sort of intuitive grasp of physics, much like how people perceive {that a} glass teetering on the sting of a desk is about to fall, or that they need to duck to keep away from a falling object.
This capability to simulate coherent, bodily believable environments over time makes Genie 3 far more than a generative mannequin. It turns into a great coaching floor for general-purpose brokers. Not solely can it generate infinite, numerous worlds to discover, nevertheless it additionally has the potential to push brokers to their limits – forcing them to adapt, battle, and study from their very own expertise in a manner that mirrors how people study in the actual world.
At present, the vary of actions an agent can take continues to be restricted. For instance, the promptable world occasions enable for a variety of environmental interventions, however they’re not essentially carried out by the agent itself. Equally, it’s nonetheless troublesome to precisely mannequin advanced interactions between a number of impartial brokers in a shared setting. Genie 3 may also solely assist a couple of minutes of steady interplay, when hours could be mandatory for correct coaching.
Nonetheless, Genie 3 presents a compelling step ahead in educating brokers to transcend reacting to inputs to allow them to plan, discover, hunt down uncertainty, and enhance via trial and error – the sort of self-driven, embodied studying that’s key in shifting in the direction of normal intelligence.
“We haven’t actually had a Transfer 37 second for embodied brokers but, the place they’ll really take novel actions in the actual world,” Parker-Holder mentioned, referring to the legendary second within the 2016 sport of Go between DeepMind’s AI agent AlphaGo and world champion Lee Sedol, wherein Alpha Go performed an unconventional and sensible transfer that turned symbolic of AI’s capability to find new methods past human understanding.
“However now, we are able to doubtlessly usher in a brand new period,” he mentioned.