At the 2026 World Robot Conference Forum held in Beijing from August 19-22, Jia Kui, founder and CEO of Cross-dimensional Intelligence, shared his perspective on how world models will serve humanity. He noted that human needs generally fall into two major categories: one is content consumption, where people enjoy the fruits of digital creativity; the other is physical labor, where there is an expectation for robots to take on heavy and demanding work. Correspondingly, the AI field has long followed two parallel tracks: perceptual understanding and reconstruction generation.

Jia Kui elaborated that if a unified world model can be developed in the future, it could generate virtual worlds with fully self-consistent three-dimensional geometry and physical laws. When this world is projected into text, it produces language generation; when projected into images from any angle, it outputs picture and video content, thereby meeting people's content needs. Simultaneously, by leveraging the world model for environmental perception and logical reasoning, it can directly drive robots and various physical agents to execute tasks, liberating humanity from labor.

In his view, the world model represents the unified hub of next-generation artificial intelligence. Once a model can precisely perceive, deduce, and reconstruct the physical world, it holds dual value: it can produce high-quality digital content while also controlling physical robots for practical operations. It can enter industrial scenarios to take on repetitive and high-risk jobs, and it can also step into homes, understanding human intentions and completing household tasks based on the characteristics of different objects.

From a scientific research perspective, Jia Kui believes the world model will become a long-term, core, and unified research topic in the fields of machine learning and statistics over the next decade.