Computer scientist at Stanford University. She is leading an approach to artificial intelligence she calls "spatial intelligence", which holds that world models must be interactive, multimodal (capable of interpreting prompts) and consistent. In her view, video-based AI systems can clear the first two hurdles but struggle with the third.
Dr Li founded World Labs, a startup that has built a world model called Marble, capable of creating digital versions of 3D worlds which are internally consistent and complete.
She is sceptical of the notion that large language models contain genuine world models. LLMs, she argues, are just "wordsmiths in the dark". Being able to use language to describe the world, she says, does not mean they have a grounded understanding of it.
The trouble with a kitten is that When it grows up, it's always a cat