As ChatGPT (ChatGPT) became a hot topic, interest in super-giant AI is increasing. Domestic companies are also paying attention to this and are moving to develop related technologies and AI semiconductors.
The National Academy of Engineering of Korea (hereinfollowing referred to as the Academy of Engineering) held a joint forum online on the 10th under the theme of ‘AI semiconductor technology to open the era of super-giant AI in Korea’.
Jeonghee Song, Executive Vice President of the National Academy of Engineering of Korea
In his congratulatory speech, Song Jeong-hee, executive vice president of the National Academy of Sciences, mentioned ChatGPT, which has recently become a hot topic, and said, “It is an event in which the existing search business model collapses.”
“Regardless of the field, interest in AI is increasing. It is a signal that we must prepare for AI that fundamentally shakes the concept of labor.”
Professor Eun-Ok Baek of Hanyang University (Chairman of the Computing Subcommittee of the National Academy of Sciences) began by saying, “The late Professor Sang-yeol Min, who achieved world-class research achievements in the field of flash memory, said, ‘Harmonious development of hardware and software is important.’”
“You have to have a big eye to see the whole as a system. I hope that the super-giant AI craze will serve as an opportunity to awaken new perspectives and critical awareness in the development of AI semiconductors.”
Naver Lee Dong-soo AI LAB Director (Photo=Captured from a broadcast video of the National Academy of Engineering of Korea)
Lee Dong-soo, director of Naver AI Lab (LAB), who gave a presentation on the topic of ‘super-large generative AI service and memory’ on this day, cited ‘cost’ as a problem with generative AI service. This giant AI cost 200 times more than a regular search.
He said that AI lightweight algorithms are important to solve these cost problems, and that he is working with Samsung to develop AI semiconductors that combine software and hardware.
Director Lee said, “We need to make lightweight AI services and hardware convergence,” adding Naver’s service experience and understanding of workloads, the latest AI model technology, Samsung’s hardware infrastructure experience, and the latest memory technology to optimize for ultra-large AI. He added that he is developing an AI semiconductor solution and expects to show it within this year.