Has the world gone beyond AI? A new storm is coming in the tech space

Have AI technologies like ChatGPT, Gemini and Cloud reached their limits? World-renowned AI scientist Yann LeCun believes that existing large language models cannot develop human-like understanding. He says that these systems respond only on the basis of pre-existing information, but are not able to understand the real world and take correct decisions on it. To overcome this shortcoming, work is now being done on a new AI technology, which is being called “World Model” and “JEPA”. Experts believe that this technology can make robots and artificial intelligence more intelligent in the future.

Why did Yann LeCun say that the current AI is not ‘smart’?

Yann LeCun, a veteran scientist in the AI ​​field, says that large language models like ChatGPT, Cloud and Gemini do many tasks well, but they cannot be called truly smart. According to him, these models only repeat the information already learned and create answers based on it. They do not have the ability to understand the real world or analyze the hidden reasons behind a situation. This is why their limitations are clearly visible in complex and real situations.

What is Zepa, on which AMI Labs is working?

Yann LeCun, former Chief AI Scientist of Meta, founded Advanced Machine Intelligence Labs i.e. AMI Labs in 2025. His company is developing a new AI technology named “Joint Embedding Predictive Architecture” i.e. JEPA. The purpose of this model is not just to provide answers, but to understand the real world and correctly assess different situations. In view of this possibility, many investors including Nvidia and Jeff Bezos’s personal investment fund have made an initial investment of more than one billion dollars in the company.

How will AI change the world model?

Yann LeCun says that humans evaluate any situation not only on the basis of memorized information, but on the basis of their experience and understanding. World Model technology works on this thinking. It creates a mental model of the real world and based on that estimates possible outcomes. For example, if a pen is held on its tip, a person knows that it will fall, but cannot tell in which direction it will fall. The World Model tries to better understand such uncertain situations.

Why is new AI technology necessary in robotics?

Today, billions of dollars are being spent around the world on creating human-like robots, but teaching them simple household tasks is still proving to be extremely difficult and expensive. Yann LeCun believes that existing large language models are almost useless for robotics, because they only understand language, not the real world. He says that superhuman intelligence cannot be achieved merely by increasing the size of LLM. A new type of AI will be required to make robots work safely and intelligently.

Many big companies of the world are working in this direction

Not only AMI Labs, but organizations like Oxford University, Google DeepMind, Wave and World Labs are also working on World Model technology. Oxford professor Ingmar Posner says that future AI systems will have to understand what is the cause of an incident and what will be the consequences of taking a different decision. Google’s Dreamer model showed its potential in this direction last year by learning how to collect diamonds while imagining future possibilities in the Minecraft game.

Will AI replace humans in the future?

Yann LeCun believes that in the future AI will first be used in the industrial sector and will gradually be prepared for general use as well. However, he says that the role of humans will not end. According to him, even if future AI systems become more capable than humans, it will still be humans who will decide what to create, which questions to ask and in which direction to move forward. He has compared future AI to an assistant that may be more intelligent than its owner, but will do the work on his instructions.

Also read: What is there in F-35 for which Türkiye is going to betray Russia?