A KAIST research team led by Professor Sungjin Ahn from the School of Computing has developed a next-generation world model that learns executable theories from observation alone. KAIST announced on the 15th of July that the team proposed a new learning paradigm called Learning-to-Theorize (L2T), which trains AI to theorize how the world works using only observed information. The team also built the Neural Theorizer (NEO), a neural network-based model that implements this paradigm. The research was selected for an oral presentation at the 43rd International Conference on Machine Learning (ICML 2026), held in Seoul from July 6 to 11, and was presented on July 9, placing it among the top 0.7 percent (168 papers) of the 23,918 total submissions. The paper was also selected for the Best Paper Award at the Compositional Learning Workshop. The L2T framework provides no predetermined answers or rules; given only a "before" and "after" observation, the AI discovers which rule produced the change. NEO discovers reusable primitives hidden within observed transformations and composes them into executable programs, such as independently learning primitives for rotation, movement, or coloring, and recombining them to explain new situations.