South Korea’s Ambitious AI Initiative: Aiming to Compete with Global Giants
In a bold move to establish itself as a leader in artificial intelligence (AI), South Korea has launched an extensive initiative to develop large language models (LLMs) tailored specifically to its language and cultural context. This initiative, which involves a significant investment of ₩530 billion (approximately $390 million), aims to reduce the nation’s reliance on foreign AI technologies and enhance national security by maintaining tighter control over data in the AI era.
A Strategic Shift in AI Development
The South Korean government, through its Ministry of Science and ICT, has selected five local companies to spearhead this ambitious project: LG AI Research, SK Telecom, Naver Cloud, NC AI, and the startup Upstage. This initiative marks a pivotal moment in South Korea’s technological landscape, as it seeks to cultivate homegrown solutions that can compete with established global players like OpenAI and Google.
The government plans to conduct biannual reviews of the participating companies, assessing their progress and potentially cutting underperformers. Ultimately, the goal is to narrow the field down to two frontrunners that will lead the nation’s AI development efforts.
LG AI Research: Innovating with Exaone
LG AI Research, the research and development arm of the LG Group, is at the forefront of this initiative with its Exaone 4.0 model. This hybrid reasoning AI model combines advanced language processing capabilities with sophisticated reasoning features, building on the foundation laid by its predecessor, Exaone Deep.
Exaone 4.0 has already demonstrated competitive performance on the Artificial Analysis Intelligence Index benchmark, alongside Upstage’s Solar Pro2. However, LG’s strategy focuses not merely on scale but on the quality of data. By leveraging real-world industry data from sectors such as biotechnology and advanced materials, LG aims to refine its models for practical applications. Co-head Honglak Lee emphasized that the company’s approach is about delivering tangible value rather than merely competing on size.
SK Telecom: Bridging AI and Everyday Life
SK Telecom (SKT) has also made significant strides in the AI landscape with its personal AI agent, A. (pronounced A-dot), which was launched in late 2023. The company recently introduced its large language model, A.X, which boasts two versions: a robust 72-billion-parameter model and a lighter 7-billion-parameter variant.
SKT claims that A.X processes Korean inputs approximately 33% more efficiently than OpenAI’s GPT-4, highlighting its advantage in local language processing. The A. service has already attracted around 10 million subscribers, offering features such as AI-generated call summaries and notes.
The company’s unique position as a telecom provider allows it to integrate AI into various aspects of daily life, from customer service to mobility solutions. Taeyoon Kim, head of the foundation model office at SK Telecom, stated that the company aims to act as a bridge between cutting-edge AI research and real-world applications, leveraging its extensive user base and telecom infrastructure.
Naver Cloud: A Full-Stack AI Solution
Naver Cloud, the cloud services division of South Korea’s leading internet company, has also made significant contributions to the AI landscape. The company introduced its large language model, HyperCLOVA, in 2021, followed by an upgraded version, HyperCLOVA X, in 2023. This model serves as the backbone for various applications, including an AI chatbot and a generative AI-driven search engine.
Naver Cloud distinguishes itself by claiming to have an “AI full stack,” which encompasses everything from data centers to consumer services. This comprehensive approach allows Naver to embed AI into its core offerings, such as search, shopping, and finance. The company believes that its real-world data and sophisticated model development will enable it to compete effectively against global giants like OpenAI and Google.
Upstage: The Startup Disruptor
Among the selected companies, Upstage stands out as the only startup. Its Solar Pro 2 model, launched in July 2023, has gained recognition as a frontier model by Artificial Analysis, placing it in competition with established players like OpenAI and Google. Despite having only 31 billion parameters, Solar Pro 2 has outperformed many global models on key Korean benchmarks.
Upstage’s strategy focuses on delivering real business impact rather than merely achieving high benchmark scores. The company is developing specialized models for various industries, including finance, law, and medicine, while fostering a Korean AI ecosystem led by innovative startups. Executive Vice President Soon-il Kwon emphasized the importance of achieving a Korean language performance that exceeds global standards.
The Broader Implications of South Korea’s AI Initiative
South Korea’s commitment to developing its own AI technologies reflects a broader trend among nations seeking to bolster their technological sovereignty. As countries around the world grapple with the implications of AI on national security, economic competitiveness, and data privacy, South Korea’s initiative serves as a case study in how to navigate these challenges.
Historically, South Korea has been a leader in technology and innovation, with companies like Samsung and LG setting global standards in consumer electronics. The current push into AI represents a natural evolution of this legacy, as the nation seeks to harness the power of AI to drive economic growth and enhance its global standing.
Conclusion
As South Korea embarks on this ambitious journey to develop homegrown AI technologies, the stakes are high. The nation’s ability to cultivate competitive large language models tailored to its unique language and cultural context could redefine its position in the global tech landscape. With significant investments and a strategic focus on collaboration among local companies, South Korea is poised to challenge established giants like OpenAI and Google, potentially reshaping the future of AI development.
