KAIST is taking the lead in developing core AI technologies through national research and development (R&D) projects such as the "Generative Artificial Intelligence (AI) National R&D Program" overseen by the Ministry of Science and ICT.
On August 13, KAIST announced that it has been selected as a joint research institution for two industry-led projects and one institution-led project under the "Generative AI Leading Talent Development Program."
The budget for each project under the Generative AI Leading Talent Development Program is 6.7 billion KRW, with the program running until the end of 2028.
First, Professor Park Noseong from the School of Computing will participate as the principal investigator for KAIST, collaborating with LG AI Research to conduct research in the field of "Physical AI." This project aims to develop image and video generation technologies and world models based on the laws of physics.
Previously, Professor Park Noseong's research team, together with Professor Yoon Sungyi's team, proposed a model architecture designed to enable AI to more precisely learn the actual rules of the physical world. This is regarded as a core technology for physics-based generative AI (Physical AI).
In a separate collaboration with NC AI, Professor Kim Taekyun from the School of Computing will serve as the principal investigator to develop multimodal AI agent technologies.
The main focus is to research technologies applicable across the game industry, including 3D modeling, animation, avatar facial expression generation, and character AI. Professor Kim Taekyun plans to contribute to nurturing practical AI talent by streamlining the game production pipeline and providing students with hands-on industry experience.
Professor Kim Taekyun, an expert in 3D computer vision and generative AI who is developing core technologies for avatar implementation in virtual worlds and the gaming industry, previously co-developed a first-person full-body motion diffusion model with Meta through joint research. He plans to apply this model to VR and AR environments.
In the institution-led project by the Korea Electronics Technology Institute (KETI), Professor Kim Seungyong from the Kim Jaechul Graduate School of AI at KAIST will participate in developing generative AI technologies. Professor Kim Seungyong's research team recently collaborated with Adobe Research and Google DeepMind to develop technology that extracts robust point tracking information from video data, presenting a core technology for clearly understanding and generating video content.
Each industry partner will jointly offer courses with KAIST and provide their own generative AI foundation models for educational and research purposes. Students selected to participate in these courses will be dispatched to industry partners to conduct practical, industry-focused research, while KAIST faculty members will also serve as adjunct professors at the dedicated AI graduate school established by LG AI Research.
Separately, KAIST faculty members are participating as key researchers in four out of the five consortia in the Ministry of Science and ICT's "Independent AI Foundation Model Development" project, solidifying KAIST's role as the central hub of generative AI research in Korea. Each consortium in this project will receive government support, including GPU infrastructure, amounting to approximately 200 billion KRW.
Lee Kwanghyung, President of KAIST, stated, "For AI technology to go beyond academic achievements and find practical applications in industry, continuous government support and research and education centered on industry-academia collaboration are essential." He added, "KAIST will continue to strive to solve real-world industrial problems and enhance the competitiveness of the AI ecosystem."
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
