Yann LeCun in Seoul: Next-Generation AI Must Understand the Real World

IT Home, October 12: According to a report today by the Korean media outlet The Korea Herald, AI pioneer and AMI Labs founder and chairman Yann LeCun said in Seoul on Monday that AI's next breakthrough will not come from large language models, but from systems capable of understanding and predicting the physical world. He believes that 'world models' will drive AI into its next stage. LeCun emphasized that large language models built for everyday scenarios have limitations in understanding the real world. He explained: 'We have not yet made robots that can complete tasks without any training, nor have we developed L5 autonomous vehicles, whereas any teenager can learn to drive after just a few hours of practice.' This shows a huge gap between AI's understanding of language and its understanding of the real world. It turns out that it is easy for machines to read language, but understanding the real world is not easy. Large language models generate text by predicting the next most likely word based on patterns learned from massive training data. LeCun believes this approach performs excellently in tasks involving predictable sequences, such as programming and mathematics, but most information in the real world is far less predictable than language. Therefore, he advocates developing world models so that AI can learn the physical laws and causal relationships governing the real world, allowing the system to predict the possible outcomes of actions and interact with its surroundings. IT Home note: LeCun is one of the important inventors of convolutional neural networks, and his theories greatly advanced modern computer vision, widely applied in image and visual recognition. He won the Turing Award in 2018 and is hailed by the industry as one of the 'godfathers of AI.' The above information comes from the original source and was organized by AI.