Google is now developing an AI “world model”
Google is developing a “world model” and universal assistant, as announced at its I/O 2025 event in Silicon Valley.
Google’s goal is to create an AI that understands the physical world and can reason and act on behalf of users, with CEO Sundar Pichai stating the company is processing 480 trillion tokens a month, a 50-fold increase from last year.
Demis Hassabis, CEO of Google DeepMind, explained that the “world model” is a model that can make plans and imagine new experiences by simulating aspects of the world, similar to the brain, and that Google is working to extend its Gemini model to become a “world model”.
The company is also developing a universal AI assistant, with Hassabis stating that the Gemini app will be transformed into a universal AI assistant that is personal, proactive, and powerful, and a key milestone on the road to artificial general intelligence (AGI).
Google demonstrated new apps and technologies at I/O, including Flow, a drag-and-drop filmmaking canvas, and Veo 3, a model that layers physics-aware video and native audio, with Hassabis highlighting that “world-model understanding is already leaking into creative tooling”.
CEO Sundar Pichai reinforced the vision, citing Project Astra, which explores the future capabilities of a universal AI assistant that can understand the world around the user, and Josh Woodward, who leads Google Labs and the Gemini App, detailed the app’s goal to be the “most personal, proactive, and powerful AI assistant”.
Google is releasing new developer tools, including Gemini 2.5 Pro with “Deep Think” and the hyper-efficient 2.5 Flash, and is making efforts to provide developer access to its technologies, such as Project Mariner, which will be released via the Gemini API.
The company’s strategy is driven by the need to defend its search business and compete with rivals such as Microsoft and OpenAI, with the goal of redefining the game with a fundamentally superior, AI-native interaction paradigm.
Microsoft is advancing its enterprise march, with a focus on Copilot and Azure AI Foundry, while OpenAI is ahead in consumer reach with ChatGPT and is planning a search offering and ad offering, posing a potential threat to Google’s search model.
Enterprise leaders are presented with compelling opportunities and critical considerations, including the need to move quickly to avoid costly rewrites, tap into revolutionary potential, prepare for a new interaction paradigm, and factor in the long game and its risks.
Google’s audacious bid to build the foundational intelligence for the AI age presents a complex and challenging landscape for the company, with the next few years being pivotal in determining its success.