The evolution of 'world models' has reached a pivotal inflection point: the shift from predicting pixels in 2D video to reconstructing the physical geometry of a dynamic world in 4D. As highlighted in Part 4 of our series for The Sequence, this frontier is defined by Spatial Intelligence'the ability of an AI to not only see a scene but to perceive its volume, its occluded parts, and its temporal trajectory with mathematical precision.
learn more
