Liberty Times (LT): You have said that Taiwan lags behind other nations in AI research and development. What specifically is Taiwan behind in and how can the nation catch up?
Chen Liang-gee (陳良基): For Taiwan to conduct AI research and develop the industry there are two main factors to establish.
First, we must have enough computational power to process research ideas. Take as an example AlphaGo, which was developed by Google DeepMind and has won multiple go games against masters worldwide. One of the program’s key developers, professor Aja Huang (黃士傑), said the program is able to make moves rapidly because of Google’s tensor processing unit, which aids the program’s rapid learning and development so it can make moves based on acquired knowledge.
For AI algorithms to be effective, they must have fast processing power to compute input data and arrive at results.
AI is a trending development across the globe and Taiwan has an advantage in development because of its strong semi-conductor industry, which offers easier access to equipment and computer chips.
Tu Yi-chin (杜奕瑾), the creator of the Professional Technology Temple (PTT) bulletin board, has said that Taiwan has many professors who have published papers on algorithmic programming and have good ideas for research.
I have criticized Taiwan for lagging behind because we have the talent and the technology, but we have yet to implement these into AI research and industrial sectors. The Ministry of Science and Technology must establish such an environment and encourage AI research and development in Taiwan.
LT: How should Taiwan link its advantages with AI research and development? How should it structure its supply line?
Chen: The ministry has planned the development and research of Taiwan’s AI industries in two areas — the semi-conductor industry and the availability of academics focused on algorithmic studies.
The main goal is not to mass-produce walking robots; that is at the end of the applicable use of AI research. The important part is the establishment of the basic, fundamental infrastructure, with an emphasis on middleware.
The government has plans to develop cloud technology, computational research and high-end processing chips. Taiwan has the best algorithms and is able to produce high-speed processing chips.
We must also provide the platforms — computers — designed for the computation. These computers must have appropriate user interface design, allowing interaction with the machine through music, video, graphics or the spoken word.
This type of sensor technology is one of the targeted strengths that Taiwan could aim toward acquiring.
The ministry plans to establish a NT$5 billion (US$164.2 million) AI creative and innovative center, that would hopefully become the main location for fostering talented individuals for AI research and development.
We also hope that professors in the algorithmic programming or other fields of AI research could attract talent to Taiwan by designing courses for basic establishment of AI technology, advanced learning of algorithmic programing and machine-training or even interface design.
These people could help facilitate upgrades to intelligent systems in companies, regardless of industry or trade, as well as develop new services or products that are more in line with intelligent systems.
RSS