Introduction to AI Breakthroughs
Over the past decade, significant foundations have been laid for the modern AI era. This includes pioneering the Transformer architecture, which serves as the basis for all large language models, and developing agent systems that can learn and plan like AlphaGo and AlphaZero.

Applications of AI Techniques
These techniques have been applied to achieve breakthroughs in various fields, including quantum computing, mathematics, life sciences, and algorithmic discovery. The goal is to continue advancing fundamental research to invent the next big breakthroughs necessary for artificial general intelligence (AGI).

Extending Multimodal Foundation Models
To achieve this, efforts are being made to extend the best multimodal foundation model, Gemini 2.5 Pro, to become a “world model” that can make plans and imagine new experiences by understanding and simulating aspects of the world, similar to how the brain functions.

Progress Towards World Models
Progress has been made in this direction, including pioneering work training agents to master complex games like Go and StarCraft, and building Genie 2, which can generate 3D simulated environments from a single image prompt.

Emerging Capabilities
Evidence of these capabilities is already emerging in Gemini‘s ability to use world knowledge and reasoning to represent and simulate natural environments, Veo‘s deep understanding of intuitive physics, and Gemini Robotics‘ ability to teach robots to grasp, follow instructions, and adjust on the fly.

Towards Universal AI Assistants
Making Gemini a world model is a critical step in developing a new, more general, and more useful kind of AI — a universal AI assistant. This AI would be intelligent, understand context, and be able to plan and take action on behalf of users, across any device.