We have a big change in our lives with the first AI …
Competition for the preoccupation of AI among domestic and foreign big tech companies is hot. The super-large AI refers to ‘AI Super Computer’, which uses AI technology to a large hardware of supercomputer class. The parameter (parameter), which is the place where the data learned by AI, usually exceeds one trillion. Parameters play a similar role to synapses that learn and remember information in the human brain.
It was a general-purpose AI specialized in language processing by the ‘GPT-3’ released by Open AI in May 2020. I learned in advance with 175 billion parameters and 300 billion datasets. Following the open AI, Google unveiled the ‘AI Switch Transformers’ with a parameter of up to 1.6 trillion in February, while Microsoft (MS) and NVIDIA last October last year’s parameter 530 billion language model’ MT-NLG ‘was introduced.
Domestic big techs have also entered the competition for establishing the AI in succession. The first place to jump is Never. Never released a 20.4 billion parameters’ hyperbola ‘Hyper Clove’, which is optimized for Korean in May last year. In December of the same year, LG introduced the first AI called ‘Exa1’ of 300 billion parameters. LG is increasing the number of ‘Exa1’ parameters to 600 billion, and is also planning a unit. Aka also introduced ‘KEPT’, a Korean-specific language AI last November. Since then, SK Telecom has developed a Korean-specific model based on GPT-3 in May, and KT plans to join hands with the Korea Electronics and Telecommunications Research Institute (TRI).
How did the AI change our lives, and what changes in the future? The market for hearing this will be held in Hall B, CODEX, Seoul on the afternoon of 15th. Sung Nacho Clove, CIC, who is in charge of the first AI in Never, Korea’s No. 1 supervisor, will give a lecture on the theme of the Tech Future Conference this afternoon. The event was organized and organized by Genet Korea in collaboration with the Korea SW Industry Association, the Korea Information and Communications Broadcasting Federation (ICT Federation), and the Digital Healthcare Forum.
Prior to the 15th, the leader of Sung Nacho, who had a preliminary interview with GAINED Korea, said, The methodology of advancing AI was roughly caught. You can see a cotton frustration.
Sex Leader, a responsible leader, majored in computer science at Seoul National University through Seoul Science High School. Since high school, he has been interested in AI, such as autonomous driving, and has developed a game engine after starting a hex flex after graduating. He worked for 18 years in the game industry, including NC soft, through Red Duck. When he was in NC soft, he developed multiplayer online games using Unrealengine3 for the first time in the world. Since June 2017, he has been a leader in Never. Below is the answer with the leader of Sung Nacho.
-The position is the leader of Never Clove CIC. What role does the responsible leader play
CIC is a company that operates as an independent company in Never, a company in the company. Never has been operating CIC since 2015. There are eight CIC’s including Search CIC’s in addition to Clove CIC. It is a business executive of the company. Never has a simple position system. There is a C-level management leader, and the CIC representative below, there is a leader of the responsible leader. It seems to be dawn.
-What do Clove CIC do? What is the organization composition?
AI calls for the AI call and AI to take care of the elderly living al1. He is also in charge of AI technologies for the Never Smart Store. While focusing. On the other hand, the AI rap takes a longer time. The staff are hundreds of clove CIC’s.
-Why did Never made a big AI?
Never has a lot of Prudent (service). Previously, AI people were deployed in each of them, and the AI personnel were distributed. We (Never) were not satisfied and customers were not satisfied. While doing, GPT3 came out, and this seemed to be the answer, so since the GPT3, Never has also launched a big AI worldwide. In order to use, there is a cost problem (AI model), which is trying to lighten the weight of the parameters. Memory CD. Ram (RAM) and other things just go out as they come in. On the other hand, the parameter digests once and sends out the result.
-What did you have been doing with a super large AI?
In the meantime, ‘hyperbola’ has been applied to Never’s representative services such as search and shopping, providing users with improved service experience. What we expect is not what you can’t do before. There was a lot of AI engineer or AI data. It was ‘hyperbola’ that made it to be d1. It’s been quite a while since the AI Care Call Service has been released in the market, and users are thankful. ‘AI Care Call Service’ It is an evolved AI service. I don’t think there is such a service in the world. Studio is an SW tool that allows anyone to create AI, such as startups and universities.
-‘Clove Studio’ is likely to contribute to the popularization of AI. Tell me more about what tools it is
No-Code is an AI platform. No code is a technology that allows you to create apps or programs with intuitive inputs such as voice or click without coding. I added the ‘tuning’ function. The newly added ‘tuning’ function is a ‘hyperbola’ language model parameter after learning user data.
It is optimized for some tasks, language, and data. The business person, the planner, and the literary students can bring AI to will and use it.
-There is a competition for preoccupying AI in the world. Why is it important?
First, I want to say AI sovereignty. The language on the Internet is 60%. Korean is only 0.6%. The language used by the US GPT-3 is 93%. It’s all about the intellectual activities we think of in language. If this goes up, the productivity of our intellectual activities can be increased. Korea’s productivity is falling behind, so I thought that Korea should give a big AI, and we opened our platform so that others would use it. AI will continue to develop. When I was crowded, I made World in Korea and made a lot of good services, but now Facebook has taken control. I look forward to the service.
-There is also a view of waste about the super large AI…
It was when I first came out of the first time. When Burt first came out, I told you where so big was used. It seems that the standard of criticism is not good. The reason for understanding is too expensive. One. When people use it, they are valuable. It is important to create this.
-What do you hear in the 15th lecture
The methodology of advancing AI seems to be roughly caught. It’s a step to think about what you can do with AI, and I’m going to talk about the strategies for doing this. If you don’t have a good AI team right now, you can see a frustration.
-The government declares an AI powerhouse and is implementing various policies. If you give advice on AI powerhouse
I often say AI these days. The data is intelligent. I can see after the data that has entered the learning. It’s a Chalmers, but it’s a lot of people who are coded by squeezing the code well. The adoption rate is half. What would happen? I thought it was impossible last year. There is a disadvantage that the parameter is too much, and if Korea takes a specialized and concentrated strategy, there is no world-class level. Of course, my thoughts may be different from other people. This is not a stage to talk about (bubbles), but you have to think about how to use it. But there is no one to teach this. If AI is AI, you only learn AI theory itself. This is AI for AI. You have to talk about AI for AI. This is an AI powerhouse. It’s something wrong. You have to do AI for AI services.