For three years, AI got smarter by reading more. Now the labs are teaching it physics.
The entire boom was built on language models. Feed them enough text and they learn to write, code and even reason. But they still don’t understand that if you push a glass off a table, it will break. They are disembodied brains that only know the world through words.
That is changing. Fast.
The shift is from ‘Language Models’ to ‘World Models’. These systems don’t just predict the next word; they predict the next physical state.
Fei-Fei Li (the ‘Godmother of AI’) just raised $230m for World Labs to build spatial intelligence. Yann LeCun at Meta argues that world models are the only path to Artificial General Intelligence. DeepMind has been quietly publishing papers on AI that understands intuitive physics.
Why does this matter? Because a language model can write a recipe, but a world model is required to build a robot that can safely cook it in your kitchen.
We are moving from AI that thinks, to AI that interacts.
