Chien Li-feng: In the AI era, the real “oil” is data
As generative AI gains momentum, businesses and industries are becoming aware of the power of AI. In light of this, the National Science Council held the “2024 GenAI Industrial Summit – AI Generation, Infinite Possibilities” on March 13 as the first event of the “Crystal Taiwan” program. Experts from various industries, academia, and government were invited to share the future possibilities of AI.
The GenAI Crystal Taiwan Industrial Summit invited experts from Taiwan’s industry, academia, and research to explore how Taiwan can use generative AI to drive technological innovation and industrial transformation across all sectors.
Chien Li-feng, former general manager of Google Taiwan, pointed out that Taiwan’s problem lies in its high concentration in a single industry (semiconductors). “We need to think about how to apply semiconductors in AI more effectively. The logic of innovation across all sectors is not about inventing new tools, but about how to utilize existing AI models to create an application model that belongs to Taiwan.”
Chien further explained, “Taiwan produces the most GPUs, but we are one of the countries that use them the least. The market value of TSMC in a year is equivalent to Taiwan’s GDP for a year. This is one of the factors that ranks Taiwan as one of the world’s most vulnerable countries, not due to cross-strait relations, but due to Taiwan’s heavy reliance on a single industry.”
He emphasized the need to think about what Taiwan should do in the next step of the industrial cycle.
“In the past financial crises, the United States printed out a China, but in today’s financial crisis, it has printed out seven data companies, which together have the same market value as China.” This data illustrates the future development of AI, Chien said.
Chien pointed out that TSMC is currently among the top ten companies in terms of market value, but one of the top ten is an Arab oil company, while the other seven are data companies. “In the AI era, the real ‘oil’ is data, and semiconductors are the process. In this process, how do we connect semiconductors with data? This is the importance of this plan!”
According to a Bloomberg report in February, Taiwan ranked ahead of China as a beneficiary of the AI era. Chien explained that this is because generative AI has struck a weakness in China, not because China’s AI capabilities are weak. In the era of generative AI, the operation is based on “questions and answers.” In the internet age, it is easy to block a webpage, but if keywords like “Winnie the Pooh” need to be blocked during AI operations, it will affect the operation. Chien stated that China’s excessive intervention in selecting successful companies has limited the development of AI in China. “There can only be AI if there is openness.”
The Bloomberg report showed that Taiwan had low values in terms of application and innovation. Chien said that Taiwan is fortunate to have a foundation in semiconductors in this AI era, but innovation and application will be important areas for future development.
The National Science Council, the Ministry of Digital Development, and National Taiwan University are actively seeking opportunities for AI in Taiwan.
In this process, collaboration between the public and private sectors is necessary.
Liu Cheng-hua, director of the National Taiwan University Smart Life Innovation Center, shared that in order to accelerate the application of generative AI in industries, they will promote generative AI innovation through the “Innovation and Entrepreneurship” and “Creative Creation” competitions. They aim to introduce generative AI to the industry and the public, generate new AI application ideas through competitions, and allow industry technology teams to propose more competitive products through the competition. “Unity is strength, and this is how we can achieve this,” Liu said, hoping to speed up the effectiveness of generative AI.
Wu Cheng-chung, chairman of the National Science Council, added that the council has been closely monitoring the latest trends in AI technology. The four-year “Taiwan AI Action Plan” has cultivated over 33,000 AI application talents and achieved a 48.7% deployment of AI in enterprises. Now, the “Crystal Taiwan Program” and the “Preliminary Plan to Promote the Introduction of Generative AI in Various Industries” will further bring together cross-departmental resources, public and private sector energy to promote the introduction of generative AI across all sectors.
The preliminary plan for generative AI is expected to be implemented for one year with a budget of NT$100 million. “But this is just the beginning. We will continue to expand and innovate to help Taiwan seize the opportunities in AI application development.”
Chien Li-feng, CEO of Appier: AI assists marketing decisions and even generates marketing plans
What does the industry think about AI?
Regarding the application of AI in the marketing field, Chien Chih-han, CEO and co-founder of Appier, mentioned that there are many areas in marketing that can be optimized. Through AI, marketers can effectively reach their target audience with their products. AI can assist marketers in making decisions, generating content that the audience wants to see, and even help create marketing plans. This simplifies the testing and communication costs for marketers, assists in interpreting marketing data, resolves the interpretation differences in marketing reports, and allows for more accurate and standardized marketing analysis reports.
Yeh Chia-shun, AI and Data Engineering Associate Manager at MediaTek, proposed the Generative AI Service Framework (GAISF), which serves as a generator for generative AI in companies. The goal is to improve the operational modes of companies and enhance their productivity. For companies, there are three important principles.
The first is “information security.” AI applications raise the most sensitive issue for companies, which is information security. The IC design industry needs to establish information security regulations to ensure that employees can use AI applications with peace of mind.
The second point is to pass on the methods used by employees. In MediaTek or any industry, experienced mentors are becoming increasingly scarce. If experience and know-how can be digitized, it can guarantee continuous inheritance and learning.
The third point is to continuously expand these models, adding more information and keeping up with the pace of AI model updates.
Yeh stated that MediaTek has collaborated with Microsoft to develop an AI assistant platform called DaVinci, which meets the aforementioned three important criteria. This platform assists MediaTek internally by integrating various company systems, reducing the reading time of technical documents for developers, and even providing each employee with their own AI assistant to assist in scheduling meetings and sending emails.
Edited by: Lin Mei-xin
