LG AI Research to Develop Super-Giant Multimodal AI With SNU

LG AI Research hosted on the 25th April a signboard hanging ceremony for the "SNU-LG AI Research Center" ("Research Center") jointly established with the department of AI at SNU, and it is fast-tracking joint research on super-giant multimodal AI.

(left) Junho Song, professor of SNU, Gyeongmu Lee, senior professor of SNU Haecheon Choi, Vice-Chancellor of the SNU Department of Research, Kyunghoon Bae, representative of LG AI Research, Jungyun Seo, AI Talent Development Supervisor of LG AI Research, Seunghwan Kim, Vision Lab leader of LG AI Research

The Research Center researches the core technology of a super-giant multimodal transcending the limitations of the existing super-giant AI that handles and expresses flat, 2-dimensional texts and images such as 3D generation technology that allows AI to emulate the body and movements of AI humans in virtual realities.

Kyunghoon Bae, representative of LG AI Research

“With SNU holding the top global AI research capabilities, we will challenge ourselves with creating AI technologies that never existed before and become a global leader in the super-giant AI through joint research on multimodal AI. Meanwhile, we will also put our efforts into developing outstanding talents,” said Kyunghoon Bae, representative of LG AI Research.

At the ceremony held on 25th, a total of 11 faculty members including Gyeongmu Lee, Byeongtak Jang, Chief of SNU AI Research Center, and Haecheon Choi, Vice-Chancellor of the SNU Department of Research attended the ceremony, joined by LG AI research staff including Kyunghoon Bae, representative of LG AI Research, and Jungyun Seo, AI Talent Development Supervisor at LG AI Research.

“With the opening of the Center, we laid the foundation for the commercialization technology research for super-giant AI. We will closely cooperate with LG AI Research to lead global-level super-giant multimodal AI research and cultivate national AI talents,” said Haecheon Choi, vice-chancellor of the SNU Department of Research, in his speech.

Photograph of "SNU-LG AI Research Center" signboard hanging ceremony

LG AI Research and SNU has recently selected 8 joint research projects to secure core technologies for super-giant multimodal AI.

For instance, senior professor Gyeongmu Lee will be researching 3D creation technology that can forge full-body image and movements with just one image or short video images. This technology can create a powerful synergy effect when applied to AI human technology. For example, when this technology is applied to "Tilda," an AI artist by LG presented in New York Fashion Week in February, Tilda’s character can freely move around within a Metaverse.

Another joint research project will be one for the next-generation language model that emulates the human language learning model. There are already many outstanding models in the language field, but they still give answers that are out of context or situation from time to time. To overcome this limitation, LG AI Research and SNU are set to develop an AI emulating the human language learning structure that independently learns external information as well as the previously learned information to make comprehensive decisions.

Both parties in the research will also study multimodal AI that can learn and use information in various formats including tables, graphs, videos with visual and audio information, and sensory information such as temperature, as well as conventional language and image information. This project is still at the very early research stage globally and is the field that must be occupied first to secure leadership in the area of global super-giant multimodal AI.

Other than this, both parties will also carry out AI advanced technology that can be applied immediately to various industries and AI ethics bias and equity research, which is regarded as an essential part of AI commercialization. In particular, it will research a continuous learning algorithm that will secure equity and transparency so that a super AI that learns a large volume of information can learn content without bias against race, gender, age, etc.

Meanwhile, LG AI Research unveiled last December "EXAONE," the super-giant AI with the best performance in Korea that is under development with more than KRW 120 billion worth of investment in the past 3 years. This February, it also launched Expert AI Alliance as the first private super-giant AI alliance forged among different industries in collaboration with 12 companies in IT, Finance, Education, Medical, Manufacturing, Communication, etc. in Korea and overseas including Google, Woori Bank, EBS, etc.

To secure its global leadership, LG AI Research is carrying out research on super-giant multimodal AI with SNU, advanced research on the up-to-date AI for reinforced learning with Michigan University, US, and MI (Materials Informatics) source technology with Toronto University, Canada to identify new medicines and materials; thus expanding its global search networks.

[Reference] Global AI Research Network of LG AI Research

▶LG AI Research Establishes an AI Research Center With SNU to Lead the Global AI Industry(Link)