Introduction of Roles and Strengths
of Participating Companies such as Liner, SelectStar, and Krafton
The SKT consortium, which was selected as an elite team for the independent artificial intelligence (AI) foundation model by the Ministry of Science and ICT, identified five key strengths: accuracy, reliability, scalability, versatility, and efficiency.
On December 14, SK Telecom posted an article on its newsroom homepage outlining the roles and strengths of the companies participating in its consortium: Liner, SelectStar, Krafton, 42dot, and Rebellion.
Liner is responsible for enhancing the accuracy of the independent AI foundation model. Jinwoo Kim, CEO of Liner, stated, "The strength of the SKT consortium lies in its experience proven through actual services, not just theory," adding, "The structure that allows us to quickly connect key stages of AI development with our internal capabilities is a competitive advantage."
SelectStar lays the foundation for a trustworthy and reliable AI model by verifying the model’s reliability and stability, ensuring customers can use it with confidence. Kim Seyup, CEO of SelectStar, said, "An evaluation system specialized for the Korean language environment is necessary," and added, "A structure that can detect and improve errors and biases that occur in real service environments at an early stage will be a key competitive factor for AI models."
Krafton’s role is to provide global scalability based on its experience in multimodal research and development (R&D). Having operated game services for users worldwide, Krafton can provide the foundation for expanding AI models globally through its world-class AI technology and accumulated user behavior data.
Kangwook Lee, Head of AI Division at Krafton, explained, "A distinctive feature of the SKT consortium is the tightly integrated structure connecting model, data, accelerator, and service," adding, "Technical capabilities proven in real service environments support the competitiveness of our independent model."
42dot is responsible for ensuring versatility through its expertise in on-device AI technology and lightweight model optimization for vehicle environments. Kim Byungyeol, Head of the GLEO AI Group at 42dot, mentioned, "Low latency, real-time decision-making, and high stability are key values in the AI model development process."
Rebellion is tasked with strengthening infrastructure functions so that domestic models can operate efficiently in real service environments. Hongseok Kim, Head of Software Architecture at Rebellion, emphasized, "The structure that enables joint optimization of domestic models and domestic semiconductors is a unique competitive advantage."
Of the five consortiums, including the SKT consortium, that are developing independent AI foundation models, the number will be reduced to four after the first evaluation by the Ministry of Science and ICT scheduled for January 15 next year.
© The Asia Business Daily(www.asiae.co.kr). All rights reserved.
