🌍 World Models: How to Give Computers Imagination?
🌍 World Models: How to Give Computers Imagination?
Standard large language models are good at understanding context and responding to prompts based on the datasets they were trained on. However, they don't understand the laws of the physical world and lack intuition or critical thinking.
The so-called "world models" that learn in real-time from the data about the world around them might soon fill that gap. Such AI systems simulate real-world environments and learn to make decisions based on the dynamics of physics and spatial properties.
Humans learn to clear a dinner table by the age of 10 and drive a car by 17—and learn both in a matter of hours. But even the world's most advanced AI systems today, built on thousands or millions of hours of data, can't reliably operate in the physical world, says Yann LeCun, Meta AI Chief. That's what imagination—a model of the world—is for.
"A world model is your mental model of how the world behaves. You can imagine a sequence of actions you might take, and your world model will allow you to predict what the effect of the sequence of action will be on the world," explains LeCun
Who Develops Such Models?
⏺ Stanford professor Fei-Fei Li: she founded the startup World Labs in April 2024;
⏺ Nvidia, at the recent CES 2025, unveiled its Cosmos platform for creating simulations to train robots;
⏺ Meta, Google DeepMind, and other companies also work in this direction.
Yann LeCun believes we will only see serious progress in world models in 10 years.
🤖 Key Areas of Application
Primarily, world models will be useful for training robots that will be able to create their own experiences instead of just learning from photos or videos.
Still, simulations based on world models can be useful not only for training robots or autonomous vehicles but also in gaming or augmented reality.
More on the topic:
👩💻 Yann LeCun on Creating AI Models with Common Sense
