In a groundbreaking development, the Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) in Abu Dhabi has introduced PAN, an advanced AI model capable of predicting future events while maintaining human-like consistency in its responses. Unlike existing systems such as OpenAI’s Sora and Google’s Veo, which often produce visually stunning but temporally inconsistent video clips, PAN excels in preserving scene structure and object continuity over extended sequences. This innovation marks a significant leap in AI technology, addressing a critical gap in current video generation models. PAN operates through a two-stage process: first, it creates an internal representation of the environment by recognizing objects, motion, and relationships over time. Second, it translates this understanding into coherent video outputs, updating them dynamically as instructions evolve. This approach ensures stability and realism, making PAN particularly valuable for applications in robotics, autonomous systems, and industrial planning. Developed collaboratively by MBZUAI teams in Abu Dhabi, Paris, and Silicon Valley, PAN is part of a broader initiative to reduce reliance on imported technology. The model and its technical documentation are freely accessible at panworld.ai, positioning the UAE as a leader in open AI research. PAN’s ability to anticipate outcomes and understand the physical world’s dynamics represents a significant step toward AI systems that can reason and predict with genuine accuracy.
AI that predicts future? UAE’s MBZUAI develops tech that anticipates ‘what’s next’
