AI development will not boom until 2024-2025 since it has not reached a critical breakthrough, according to provost Richie Tsai for Taiwan AI Academy.

Development of AI, software and semiconductors will drive new economic development over the next 10 years, Tsai said.

Taiwan-based makers have played a globally important role in production of PCs, notebooks and smartphones, but global demand for these products has been in slow growth or even decreased, and is unlikely to rebound, Tsai noted.

Over the next 10 years, HPC (high-performance computing), AIoT equipment and devices, electric vehicles and medical devices will drive the growth in IC demand, Tsai indicated.

AI development consists of three aspects: core AI technology, including machine learning, deep learning, advanced learning, tool architecture and data governance; AI application, including computer vision, natural language processing, data analytics and prediction, recommendation system; industrial application of AI, including smart manufacturing, smart medical care, smart retail, smart finance and edge computing (AI+IoT), Tsai said.

Many enterprises begin with the industrial application of AI, but if they want to gain long-term competitive advantages, they need to understand core AI technology, Tsai noted. Viewing that a few international high-tech companies have oligopolized core AI technology, Taiwan AI Academy has offered AI curriculums in a bid to break the oligopoly, Tsai indicated.

Many Taiwanese ICT makers have entered supply chains for smart medical instruments and devices, Tsai said. Besides, they are considerably advantageous to step into edge computing, for edge devices outnumber cloud computing devices, and edge computing application is not confined to cloud servers, Tsai noted.

For the industrial application of AI, PoC (proof of concept) is essential, Tsai said. Taiwan is the world's best experimental field for PoC because Taiwan is small in land area and has high population density, high popularization of education, globally leading Internet environment and high smartphone penetration, Tsai indicated.

Particularly for industrial application, many AI-based solutions are in transition from PoC stage to commercialization, Tsai said.

Along with the development of AI, enterprises have to re-define employees' skills, functions of operational departments or business units, organizational efficiency as well as seek partners to develop ecosystems concerned, Tsai indicated.

According to semiconductor distributor WPG Holdings - an Intel IoT solution aggregator - the government sector accounts for 26% of the global volume of IoT devices, manufacturing industries for 22%, transportation industry for 15%, retail industry for 12%, medical care for 9%, logistics for 7%, agriculture for 4%, home application for 3%, others for 2%.

By DIGITIMES