Yann LeCun: AI's Next Breakthrough Will Come From World Models, Not LLMs

According to Korea's The Korea Herald, Turing Award winner and AMI Labs founder Yann LeCun said in Seoul that AI's next breakthrough won't come from large language models. Instead, it will come from systems that can understand and predict the physical world. He calls these "world models."

He believes large language models have limits in understanding the real world. Robots still can't finish tasks with zero training, and fully self-driving cars are still not here. Yet a teenager can learn to drive in just a few hours. This shows a big gap between understanding language and understanding reality.

LLMs create text by predicting the next likely word. They do well in coding and math, but real-world information is much harder to predict. LeCun suggests building world models that learn physical rules and cause and effect, so AI can predict outcomes and interact with its surroundings.