Meta has unveiled its new AI model called "V-JEPA 2," which is described as a "world model" designed to enable AI agents to better understand the physical world and operate more effectively within it. The model allows AI to observe processes around it and predict their outcomes — functioning similarly to the intuitive reasoning seen in children and animals.
For example, if an AI agent is holding a plate and a spatula while moving towards a stove with cooked eggs, V-JEPA 2 can predict that the most likely next action is to use the spatula to transfer the eggs onto the plate. These types of “everyday” actions help robots comprehend their environment and the physical laws that govern it.
According to Meta’s chief AI scientist, Yann LeCun, this model enables robots to understand their surroundings without needing to be trained on millions of hours of data. He states that such world models could usher in a new era of AI, especially in robotics.
Meta claims that V-JEPA 2 is 30 times faster than Nvidia’s Cosmos model, which also focuses on physical world intelligence. However, it should be noted that this comparison may be based on different evaluation criteria.
V-JEPA 2 represents the next significant step in AI’s ability to understand the world around it. This technology could pave the way for robots to become more integrated into our daily lives and make natural, context-aware decisions.
