Latest news with #ExaoneDeep
Korea Herald
7 days ago
- Business
- Korea Herald
Lee turns to technocrats as Korea pushes for AI leadership
Experts say Bae Kyung-hoon's nomination reflects urgency, realism in Korea's AI policy shift President Lee Jae Myung sent a clear signal Monday that South Korea is going all-in on artificial intelligence, tapping AI expert Bae Kyung-hoon, chief of LG AI Research, as the new science minister to spearhead the country's AI initiative. Monday's nomination came on the heels of another high-profile pick from the private sector. On June 15, Ha Jung-woo, the head of Naver AI Innovation Center, was appointed as the first senior secretary of AI and future planning. The newly established post is tasked with leading the country's investments and policies for AI infrastructure. Presidential Chief of Staff Kang Hoon-sik explained that the president appointed the 49-year-old AI strategist to help propel Korea into the ranks of the world's top three AI powerhouses. 'We expect Bae to strengthen AI competitiveness alongside Ha Jung-woo,' said Kang. Bringing in top experts from Korea's tech giants LG and Naver highlights how serious the Lee administration is about making Korea a global AI leader. Lee has announced plans to invest over 100 trillion won ($72.29 billion) in public-private funds to push the AI industry forward. 'With both Ha and Bae on board, it is clear that the Lee administration's focus is on AI,' said Hwang Yong-sik, a professor at Sejong University College of Business and Economics. 'There is a growing recognition that relying solely on the public sector for AI development has its limitations. The industry needs both the government and the private sector. While the government can lead, AI advancement is not possible without support from the private sector, hence the appointment of experts from companies.' Bae's career spans startups and South Korea's largest corporations, including Samsung, SK and LG. In 2006, he worked as a senior researcher at Samsung Thales, a joint venture between now-defunct Samsung Techwin and French firm Thales. From 2011 to 2016, he was at SK Telecom's future technology R&D center Bae joined LG Group in 2016 as an AI expert, taking on various AI roles at LG Economic Research Institute, LG Uplus and LG Science Park. In 2020, he became the founding president of LG AI Research, the conglomerate's dedicated AI think tank. Under Bae's leadership, LG AI Research developed the hyperscale language model Exaone in 2021, commercialized it in 2023, and released its third version as open source in 2024. Most recently, his team unveiled Korea's first interference AI model, Exaone Deep, in March. In addition to his corporate leadership, Bae has also served in key government advisory roles related to AI governance and privacy. Not many people in Korea have such deep experience and expertise in the still relatively nascent industry as Bae, said Choi Byung-ho, a professor at Korea University's Human-inspired AI Research Lab. 'In Korea, only a handful of people in the field can handle the frontier model from start to finish, and Bae is one of them," said Choi. Choi stressed the urgency of AI development in Korea. 'We are in a race against time, and speed matters more than ever," he said. "Decision-making has to be fast and precise. We can't afford to put a generalist or bureaucrat in that role. Only experts with real field experience are what's needed.' Once the parliament gives the green light, Bae will be in charge of the Ministry of Science and ICT and oversee the country's science and digital infrastructure policies. He and Ha will work closely to develop national strategies and coordinate interagency efforts to execute AI policies. Both strongly advocate developing homegrown AI models using domestic infrastructure, data and expertise. Their appointments are expected to inject fresh momentum into the administration's push for sovereign AI. At a National Assembly hearing earlier this year, Bae warned that not developing a sovereign AI foundation model would be akin to 'forfeiting national strategic assets,' calling for a strategic approach. Going forward, when asked what is most important at the government level to accelerate AI development and build a Korean model, experts consistently point to infrastructure, which includes data centers, energy systems, supercomputers, graphics processing units, software and data resources. 'Infrastructure is the foundation, and Korea doesn't have the infrastructure it needs,' said Choi. 'At a time when frontier models evolve into global models, Korea needs more infrastructure to make a global model. This is a period of explosive growth. AI develops week to week. Both short-term and long-term strategies need to happen in parallel.'
Korea Herald
18-03-2025
- Business
- Korea Herald
LG to take on OpenAI, DeepSeek with Korea's first reasoning AI model
LG AI Research has unveiled the country's first reasoning artificial intelligence model, Exaone Deep with top-tier performance to compete head-on with industry-leading players including OpenAI and DeepSeek, Tuesday. The research institute, affiliated with Korea's fourth-largest conglomerate LG, introduced the new foundation model at this year's Nvidia GTC, an AI conference held in Silicon Valley. The company claimed the model outperforms leading US and Chinese AI models with fewer parameters and graphic processing units. According to LG, Exaone Deep 32B features 32 billion parameters, which is only 5 percent of the parameters of China's DeepSeek R1 -- runs on 671 billion parameters -- but demonstrates exceptional performance, especially in solving complex mathematical and scientific problems. A parameter is a numeric value that helps AI models process data and make decisions. "Exaone Deep shows excellence in high-difficulty benchmarks, outperforming significantly larger models. In terms of the number of GPUs each model uses, Exaone Deep 32B runs on only one unit of Nvidia's H100 chip, while DeepSeek operates on 16 of those expensive chips," an LG AI research official explained, highlighting the learning efficiency and cost-effectiveness. "Our reasoning AI model shows how Korea's AI technology is globally competitive, given how only a few players -- mainly in the US and China -- have developed in-house reasoning AI." Exaone Deep 32B model demonstrated competitive superiority in science and coding domains, scoring higher than its rivals. The model achieved 94.5 points in the 2025 Korean College Scholastic Ability Test mathematics section when OpenAI o1-mini scored 84.4 points and DeepSeek R1, 88.8 points, in the data provided by the research center. On the American Invitational Mathematics Examination 2025, Exaone Deep 32B achieved performance on par with the DeepSeek R1, with both models scoring 80 points. The 32B model also achieved a 66.1 score on the Graduate-level Google-Proof Q&A Benchmark that evaluates PhD-level problem-solving abilities in physics, chemistry and biology. In LiveCodeBench, the benchmark for evaluating coding capabilities, it scored 59.5 points, outperforming other reasoning AI models of similar size, LG said. "Humans think, reason and produce results. For AI agents and robots to evolve into what humans truly want and satisfy demands to the level we want, inference capability is the enabler," an LG official said. Immediately after its release, the Exaone Deep 32B model was listed on Epoch AI's Notable AI Models, marking another achievement following Exaone 3.5 and making the brand the only Korean model to be listed in the past two years, the company said. In the unveiling, LG also introduced the lightweight model Deep 7.8B and on-device model 2.4B, which feature fewer parameters but prove their practical utility. Being 24 percent the size of the 32B model, the lightweight model maintains 95 percent of its performance, while the on-device model achieved 86 percent of the big model's performance at just 7.5 percent the size, the company said. Being Korea's first reasoning AI foundation model, Exaone Deep also delivers exceptional performance in general language understanding, securing the highest score of 83 in the Massive Multitask Language Understanding benchmark, an LG official said, noting how the lightweight 7.8B model outperformed the OpenAI o1-mini in the US. "The core of LG's AI technology is reducing model size while maintaining performance.' LG AI Research released all Exaone Deep models as open-source, allowing researchers, developers and other users to access and build on them. Meanwhile, Koo Kwang-mo, CEO of LG Corp., emphasized in his New Year's address this year, "The current LG has been built by accumulating many moments of challenging new areas and creating unprecedented value." "We will create a new lifestyle where people use cutting-edge technologies such as AI conveniently in their daily lives, allowing them to spend their precious time on more enjoyable and meaningful tasks," Koo said.
