SDF

KIM Sung

In the aftermath of the generative AI craze sparked by ChatGPT, Large Language Models (LLM) have become the newest industrial trend. Private LLMs, in particular, are being heralded as game changers in the AI market thanks to their ability to learn using only internal company data, which eliminates the risk of data breaches, and the way they have been designed to prevent hallucinations, or the generation of false information. But the LLMs of Big Tech, trained using foreign languages, are weak in their grasp of the Korean language, not to mention Korean cultural sentiments and regional information, which has been an obstacle for advancing private LLMs used by local companies. This was the reason behind Upstage’s launch of the 1T Club, the aim of which is to collect 1 trillion Korean-language tokens, or basic units of text and code, and its efforts to achieve Korean “LLM independence” through the creation of an ecosystem that enables a mutually beneficial interchange between providers of data and the companies creating models. Through “the world's best private LLM, created with the right data,” Upstage seeks to realize AI innovation that will transform Korea’s industrial landscape.