Historically, computer vision treated 3D reconstruction (capturing reality) and generation (creating content) as separate fields. New techniques like NeRFs are merging them, creating a unified approach where models can seamlessly move between perceiving and imagining 3D spaces. This represents a major paradigm shift.

AI lab General Intuition uses video game data to train AI that understands space, time, and human action. This is a richer dataset for building 'world models' than the text-based data used for LLMs, with applications far beyond the gaming industry itself.

GI discovered their world model, trained on game footage, could generate a realistic camera shake during an in-game explosion—a physical effect not part of the game's engine. This suggests the models are learning an implicit understanding of real-world physics and can generate plausible phenomena that go beyond their source material.

Inspired by LLMs, Atlas treats 3D reconstruction as "generation with a really long context." This allows the model to ingest dozens of views to ground its output, bridging the gap between purely imaginative generation and precise, data-driven reconstruction.

Large language models are insufficient for tasks requiring real-world interaction and spatial understanding, like robotics or disaster response. World models provide this missing piece by generating interactive, reason-able 3D environments. They represent a foundational shift from language-based AI to a more holistic, spatially intelligent AI.

The concept of a 'world model' is evolving from action-conditioned video predictors to single, multimodal models like Google's Omni. Omni demonstrates a deep, scalable understanding of the world, shown through nuanced video editing, representing a more practical approach than traditional, computationally expensive architectures.

For the first time, Atlas combines the traditionally separate fields of creative pixel generation (like text-to-video) and precise 3D reconstruction into one architecture. This dual capability allows it to both imagine and accurately map physical spaces.

Atlas is built on predicting the next view from any camera angle, a fundamentally different primitive than the next-token prediction of LLMs or the next-frame prediction of video models. This approach enables true spatial reasoning and understanding.

Current multimodal models shoehorn visual data into a 1D text-based sequence. True spatial intelligence is different. It requires a native 3D/4D representation to understand a world governed by physics, not just human-generated language. This is a foundational architectural shift, not an extension of LLMs.