Artificial intelligence (AI) has not been of much help tackling the coronavirus pandemic, as there is no historical data on the virus that can be used to train AI models, and data collection has been too slow to cope with the fast spread of the disease, according to Kung Hsiang-tsung, president for Taiwan AI Academy.

There are two main reasons behind AI's limitations: the length of time needed to develop AI-based solutions; and difficulties in integrating data across different fields, said Kung, who is a computer science and electrical engineering professor at Harvard University.

Development of AI-based solutions entails integration of relevant data first and then use of such data to train AI models, which is a lenghty process thta can take weeks, months or years to complete, said Kung.

Communication and coordination for data integration among different fields are complicated, Kung said.

In the case of coronavirus pandemic, AI has been used to help with vaccine R&D, but clinical trials take time and cannot catch up with spread and genetic mutation of the virus, said Kung.

Founded in 2017, Taiwan AI Academy is jointly sponsored by Formosa Plastics, Chimei, Inventec, Elan Microelectronics, MediaTek and AU Optronics (AUO). The academy aims to help firms develop problem-solving AI-based solutions.

Taiwan AI Academy president Kung Hsiang-tsung
Photo: Michale Lee, Digitimes, November 2020