NC has partnered with KAIST to conduct foundational research on AI that understands physical laws, addressing the limitation that existing generative AI fails to grasp concepts like gravity, friction, and collision. Professor Kim Tae-kyun of KAIST’s School of Computing, who joined KAIST in 2020 after completing his Ph.D. at the University of Cambridge and working at Imperial College London, is collaborating closely with NC’s AI organization. His core research focuses on embedding physical common sense into AI models, enabling AI to recognize physically implausible situations and produce more efficient and realistic results. The collaboration centers on Multimodal AI, integrating text, images, audio, video, and 3D data into a single model, with a focus on video generation leveraging NC's proprietary language model, VARCO. As part of the Institute of Information & Communications Technology Planning & Evaluation (IITP)'s Leading Talent Training Project, two core technologies are being advanced: part-based mesh generation technology, which automatically generates individual parts like the torso, arms, legs, and clothing separately before reassembling them to improve generation quality and convenience.