Demis Hassabis, CEO of Google’s AI analysis group DeepMind, appeared to recommend Tuesday night that Veo 3, Google’s photo-sharing app video-generating mannequin, might probably be used for video video games.
In response to a submit on X beseeching Google to “Let me play a online game of my veo 3 movies already,” and asking, “playable world fashions wen?” Hassabis responded, “now wouldn’t that be one thing.”
On Wednesday morning, Logan Kilpatrick, lead product for Google’s AI Studio and Gemini API, chimed in with a reply: “🤐🤐🤐🤐”
Each posts from the Google executives are little greater than playful ideas, and a Google spokesperson informed TechCrunch the corporate had nothing to share in the mean time. However constructing playable world fashions isn’t outdoors the realm of potentialities for the tech big.
World fashions are completely different from video-generation fashions. The previous simulates the dynamics of a real-world atmosphere, which lets brokers predict how the world will evolve in response to their actions. Video-gen fashions synthesize sensible video sequences.
Google has plans to show its multimodal basis mannequin, Gemini 2.5 Professional, right into a world mannequin that simulates features of the human mind. In December, DeepMind unveiled Genie 2, a mannequin that may generate an “limitless” number of playable worlds. The next month, we reported that Google was forming a brand new workforce to work on AI models that can simulate the real world.
Others are engaged on constructing world fashions — most notably, AI pioneer Fei-Fei Li. Li got here out of stealth final 12 months with World Labs, a startup that has constructed its own AI system that generates video game-like, 3D scenes from a single picture.
Veo 3, which continues to be in public preview, can create video in addition to audio to associate with clips — something from speech to soundtracks. Whereas Veo 3 creates sensible actions by simulating real-world physics, it isn’t fairly a world mannequin but. As an alternative, it could possibly be used for cinematic storytelling in video games, like cutscenes, trailers, and narrative prototyping
The mannequin can also be nonetheless a “passive output” generative mannequin, and it (or a future Veo era) would wish to shift to a simulator that’s extra lively, interactive, and predictive.
However the actual problem with online game manufacturing isn’t simply spectacular visuals; it’s real-time, constant, and controllable simulation. That’s why it would make sense to see Google take a hybrid method that leverages Veo and Genie sooner or later, ought to it pursue online game or playable world growth.
Google might discover itself competing with Microsoft, Scenario, Runway, Pika, and, eventually, OpenAI’s video-generating model Sora.
Given Google’s deliberate strikes on the earth fashions house and its repute for utilizing its deep pockets and distribution muscle to steamroll rivals, rivals on this house can be clever to maintain an in depth watch.