Seeking Big Advantage in AI, South Korean Companies Think Smaller

ChatGPT, Bard, Claude. The world’s most popular and successful chatbots are trained on data mined from vast swathes of the Internet, reflecting the cultural and linguistic dominance of the English language and Western perspectives. This has sparked concern about the lack of diversity in artificial intelligence. There are also fears that the technology will remain the preserve of a handful of American companies.

In tech powerhouse South Korea, companies are taking advantage of technology’s malleability to shape AI systems from the ground up to meet local needs. Some have trained AI models with rich data sets about Korean language and culture. South Korean companies say they are developing AI for Thai, Vietnamese and Malaysian audiences. Others are seeking clients in Brazil, Saudi Arabia and the Philippines, as well as industries such as medicine and pharmaceuticals.

This has fueled hopes that AI can become more diverse, work in more languages, be adapted to more cultures, and be developed by more countries.

“The more competition there is, the more robust the systems will be: socially acceptable, more secure and more ethical,” said Byong-Tak Zhang, a professor of computer science at Seoul National University.

While there are a few prominent non-US AI companies, such as French firm Mistral, the recent uprising of OpenAI, the maker of ChatGPT, has highlighted how concentrated the industry remains.

South Korea’s emerging AI landscape is one of the most competitive and diverse in the world, said Yong Lim, a law professor at Seoul National University who is leading its AI policy initiative. The country’s export-driven economy has encouraged new companies to look for ways to tailor AI systems to specific companies or countries.

South Korea is well-positioned to develop AI technology, developers say, given that it has one of the most connected populations in the world to generate large amounts of data to train AI systems. Its tech giants have the resources to invest heavily in research. The government was also encouraging: It provided companies with money and data that could be used to train large language models, the technology that powers AI chatbots.

Few other countries have the combination of capital and technology needed to develop a large language model capable of powering a chatbot, experts say. They estimate that it costs between $100 million and $200 million to build a base model, the technology that serves as the basis for AI chatbots.

South Korea is still months behind the United States in the AI ​​race and may never fully catch up as leading chatbots continue to improve with more resources and data.

But South Korean companies believe they can compete. Instead of going after the global market like their American competitors, companies like Naver and LG have tried to target their AI models at specific industries, cultures or languages ​​instead of relying on the whole ‘Internet.

“The localized strategy is a reasonable strategy for them,” said Sukwoong Choi, a professor of information systems at the University at Albany. “American companies are focusing on general-purpose tools. “South Korean AI companies can target a specific area. »

Outside of the United States, AI’s prowess appears to be limited in scope. In China, Baidu’s answer to ChatGPT, called Ernie, and Huawei’s big language model have seen some success at home, but they are far from dominating the global market. Governments and companies in other countries like Canada, Britain, India and Israel have also said they are developing their own AI systems, although none have yet released a system that can be used by the public.

About a year before the release of ChatGPT, Naver, which operates the most widely used search engine in South Korea, announced that it had successfully created a large language model. But the chatbot based on this model, Clova X, was only released this month. Septemberalmost a year after ChatGPT’s debut.

Nako Sung, an executive at Naver who led the company’s generative AI project, said the timing of ChatGPT’s release surprised him.

“Until then, we took a conservative approach toward AI services and explored possibilities cautiously,” Mr. Sung said. “Then we realized that the schedule had been accelerated a lot,” he added. “We decided to move immediately. »

Naver now runs an AI model designed for Korean language speakers using data from the South Korean government and its search engine, which has operated the country’s Internet since 1999.

Clova Naver’s chatbot is also integrated into the search engine, allowing users to use the tool for shopping and traveling.

Outside of its domestic market, the company is exploring business opportunities with the Saudi government. Japan could be another potential customer, experts say, since Line, a messaging service owned by Naver, is widely used there.

LG also created its own generative AI model, the type of artificial intelligence capable of creating original content based on inputs, called Exaone. Since its inception in 2021, LG has worked with publishers, research centers, pharmaceutical companies and medical companies to adapt its system to their data sets and provide them with access to its AI system.

The company targets businesses and researchers rather than the general user, said Kyunghoon Bae, director of LG AI Research. Its subsidiaries have also started using its own AI chatbots. One of the chatbots, designed to analyze chemistry research and chemical equations, has been used by researchers building new materials for batteries, chemicals and drugs.

“Rather than letting one or two best AI systems dominate, it is important to have a range of models specific to a domain, language or culture,” said Honglak Lee, chief scientist of the branch of LG AI research.

Another South Korean giant, Samsung, last month announced Samsung Gauss, a generative AI model used internally to compose emails, summarize documents and translate text. The company plans to integrate it into its mobile phones and smart home appliances.

Other major companies have also said they are developing their own large language models, making South Korea one of the few countries with so many companies building AI systems. KT, a South Korean telecommunications company, said it was working with its Thai counterpart, Jasmine Group, on a broad language model specializing in the Thai language. Kakao, which creates an eponymous super app for cats, said it was developing Generative AI for Korean, English, Japanese, Vietnamese and Malaysian.

However, the domination of the United States in the field of AI seems assured for the moment. It remains to be seen to what extent countries can catch up.

“The market is convulsing; it’s very difficult to predict what will happen,” said Mr Lim, the AI ​​policy expert. “It’s the Wild West, in a sense.”