If LLMs like ChatGPT are the hottest thing in AI, then world models are the holy grail.
Hailed as the most probable path to AI superintelligence by two of the three most influential AI researchers in history, Yann LeCun and Yoshua Bengio, they represent a vision of an AI that learns about our world not by brute force or rote memorization, like ChatGPT, but by forming abstract representations of it, just like humans.
In this divine narrative, Image-based Joint-Embedding Predictive Architecture (I-JEPA), built by Meta, emerges as the first tangible success in realizing this vision.