spatial intelligence

In AI, the capacity of a system to perceive, represent, reason about and act within three-dimensional space and the physical world — understanding geometry, objects, motion and physical cause and effect — as opposed to manipulating text. Fei-Fei Li, who popularized it as a research goal, argued that it requires world modelsworld model, n.: An AI's internal simulation of how its world works, which lets it predict what will happen next — especially in response to its own actions — and so plan, imagine or practice without acting for real. that are generative (producing consistent simulated worlds that obey geometry and physics), multimodal (able to take in any mix of text, images, video, depth maps, gestures or actions) and interactive (predicting the next state of the world given an action). It is pursued through 3D generation, scene reconstruction such as Gaussian splattingGaussian splatting, n.: A way of turning photos of a real place into a 3D scene made of millions of tiny, soft, colored blobs that can be viewed from any angle in real time, even on a phone or headset., video world models and embodied AIembodied AI, n.: AI that has a body, usually a robot and sometimes a simulated character, and learns or acts by sensing and moving in a world rather than only processing text or images., and is considered essential for robotics, design and scientific simulation.

Our chatbot can describe a kitchen perfectly, but it has no spatial intelligence about where the mug actually is.