Demis Hassabis, the visionary CEO of Google's AI research division DeepMind, recently hinted at the potential of Veo 3, Google's cutting-edge video-generating model, to revolutionize the video game industry. In a playful exchange on social media, Hassabis responded to a user's request to play a video game using Veo 3 videos, saying, "now wouldn't that be something." While Google has not confirmed any immediate plans, the idea of building playable world models is not far-fetched for the tech giant.

World models differ from video-generation models in that they simulate the dynamics of a real-world environment, allowing agents to predict how the world will evolve in response to their actions. Veo 3, which is still in public preview, can create both video and audio to accompany clips, from speech to soundtracks. It generates realistic movements by simulating real-world physics, making it a valuable tool for cinematic storytelling in games, such as cutscenes, trailers, and narrative prototyping.

However, Veo 3 is still a "passive output" generative model, and for it to be used in video game production, it would need to shift to a more active, interactive, and predictive simulator. The real challenge in video game production lies in achieving real-time, consistent, and controllable simulation.

Google could potentially take a hybrid approach, leveraging both Veo and Genie, its multimodal foundation model, to develop video games or playable worlds in the future. This would allow the company to combine the strengths of both models, creating immersive and interactive gaming experiences.

Competitors in the video-generating model space, including Microsoft, Scenario, Runway, Pika, and OpenAI's Sora, should keep a close eye on Google's moves. With its deep pockets and distribution power, Google has the potential to dominate the market and reshape the future of video games and playable world models.