In the short term, companies that make AI agents will win in the artificial intelligence (AI) market..

In the short term, companies that make AI agents will win in the artificial intelligence (AI) market. In the long run, companies that combine existing services or products well with AI will eventually dominate the market."

Jang Jin-seok, CEO of BCG's digital division, who recently interviewed Maeil Economy, advised as follows.

Partner Jang is a digital transformation expert who leads BCG Seoul Office Digital and 'BCGX', an AI organization dedicated to AI. He has experience in helping customers implement actual businesses such as mid- to long-term digital transformation strategy consulting and mobile application development. In particular, a number of productivity improvement programs using AI have been conducted. Recently, there has been a controversy over whether AI-related businesses can secure profitability compared to the high investment amount. This is because AI semiconductor companies such as Nvidia make money, but companies that have launched AI services are still having difficulty making a surplus. In response, partner Jang said, "AI service companies are still 'money-eating hippos' and are pouring a lot of money," adding, "A number of companies, including OpenAI, Google, and Purple Lexity, have invested beyond imagination in creating AI algorithms and giant language models (LLM)."

"These foundation model companies will eventually make money, but companies that directly service AI will be able to make profits quickly," he said. "Because the cost of using the giant model is falling sharply, it is possible to secure profitability based on lower procurement costs."

The following is the Q&A with Jang's partner.

-How AI Companies Look To Make Money

The current method is to create a service that can utilize ▷AI and pay customers for this service. However, in the future, the way of making money from AI will be more advanced. There will be more cases of securing profitability by combining AI with existing services or products. For example, Adobe has combined AI with services such as existing Photoshop, allowing it to evolve from a company that simply sold solutions to a subscription-type service company to hold customers. If this trend appears, it should be seen that information technology (IT) service companies will eventually evolve into AI service companies. This is because IT companies will strengthen their already existing profitability models with AI.

-What AI services are likely to be in the spotlight in the short term?

▷The AI agent should be noted. AI agents are automation tools that replace certain tasks based on Generative AI. For example, past presentations were designed by humans themselves. However, AI agents can automatically generate initial presentations. The rapporteur can modify it based on this. Right now, it will be important who creates which AI agent and is chosen by real customers. In particular, in the case of business-to-consumer (B2C) AI agents, existing services will also be selected by customers.

-What are the advantages of Korean companies in the field of AI relatively?

▷AI value chain consists of hardware, foundation model, domain-specific model, and agent. First of all, there is a question of whether Korea can make a good foundation model. This is because we are concerned about whether trillions of investments such as OpenAI, Google, and Meta Platforms will be possible. Since developing a foundation model requires enormous computing power and this is ultimately a cost, it will be a very big challenge if Korean companies jump directly.

After all, Korean companies should focus on services that are more in touch with consumers in the AI value chain. Since Korea is a manufacturing powerhouse, it should be able to create services such as AI agents well and incorporate them into the manufacturing environment.

-Opinions on the lack of LLM models based on Korean data.

▷There is a possibility that a better model will come out when a model is made with only Korean data, but the important thing is the data source. Already, there are concerns in the industry that the data is being exhausted by using all of the open data for learning. Accordingly, there is an opinion that customized LLM should be created for a specific industry or domain. AI's language ability is already equipped with current technology. In the future, it is important to acquire data on specific fields to learn specific knowledge. In other words, the outcome will be whether you can create LLMs with expertise in domains rather than languages.

In the past, AI was a lump. However, it is already divided into a foundation model and a fine tuning model. In other words, it is important to have a technology that can fine tune knowledge suitable for the purpose in language proficiency made from the foundation model. This means that opportunities have rather increased. This is because opportunities can be found even with AI optimized for a specific domain.

-Is there a case of a company that makes good use of AI technology?

▷There is global pharmaceutical company Sanofi. Sanofi has a vision to become the first AI-powered pharmaceutical company. In fact, many efforts have already been made to shorten the new drug development process through AI. Sanofi also has this vision, but it has introduced AI in more earnest. Sanofi added AI to the entire pharmaceutical supply chain, including research and development (R&D) such as the process of finding new drug candidates, regulatory reporting processes, and clinical trials. Sanofi has achieved enormous manpower and cost savings through this.

-What policy will U.S. President-elect Donald Trump implement on AI after his decision to return to power?

▷Right now, U.S. companies are leading AI technology, but the importance of 'AI sovereignty' is rising. As a result, competition between countries is also being created in AI. Many parts of the world, including Korea, China, India, and Europe, are thinking about ways to domesticate AI.

This is a good subject for Trump. Trump values his country's interests. In other words, it means that 'America's AI' can be used as a topic. There is a possibility that the U.S. will continue to be ahead in the field of AI, so it will implement a fairly AI-friendly policy. Recently, there has been talk of "Responsible AI," but we cannot rule out the possibility that Trump will consider it a significantly lower proportion and implement policies to help U.S. AI technology develop at a faster pace.

[Reporter Jonghwa Lee]