ATIP reports on developments in a number of leading-edge science and technology (S&T) fields. The rapid, worldwide response and development of artificial intelligence (AI) suggests something unique is occurring. Our analysts on the ground in Asia are viewing these dynamic developments first hand -- here is their timely story.
With expectations of AI influencing the global economy by trillions of U.S. dollars over the next 10 years, AI is rapidly transforming the technological and industrial landscapes of the world. It is also clear that hardware, software, and big data have combined to open an AI-driven Pandora’s box, and its broad and ground-breaking influences, both good and bad, have attracted significant attention from academia, public administration, and the commercial sector.
In the present AI boom, the overused term “AI” occurs very broadly. In many cases, “machine learning” would be more correct, as there is still a long way to real, human-like “intelligence.” In general, AI refers to machines and systems that mimic human-like cognitive functions such as learning, perception, problem solving, interaction, and reasoning. Certainly, AI developments over the past 10 years have added more seemingly intelligent and human-like capabilities to applications such as image or language recognition, and AI has rapidly become usable in our everyday lives.
The modern history of AI dates to the 1950s, in particular to a 1956 conference at Dartmouth College where the term “artificial intelligence” was used for the first time. Over the subsequent decades, researchers have made slow but important progress. Nevertheless, popular enthusiasm and expectations around the concept of an “intelligent” entity has often run ahead of reality. This has led to several periods of over-hype and subsequent disappointments. For example, the term “AI winter” is commonly used to describe just such a situation during the 1970s.
Starting about 20 years ago and continuing to the present, a combination of factors (e.g., computing power, algorithms, and big data) have come together to expand the types of problems that can be solved by tools associated with AI. These factors include new advances in computer hardware such as graphics processing units (GPUs) and other specialized chips, some with architectures loosely inspired by the human brain. New hardware has dramatically brought down the cost of computing so that techniques previously considered impossible or impractical are now routine. For example, a robot’s ability to sense and react to its environment is expanded by more capable onboard computing. Similarly, faster networks and the development of cloud computing have meant that very large computing infrastructures are accessible at modest cost to researchers, even in remote locations – an important democratization.
Along with new hardware, new techniques and algorithms are being developed and implemented, such as machine learning (ML) and neural networks (NNs), including deep learning (DL). These lead to innovative application ideas in areas including computer vision (CV), natural language processing (NLP), autonomous vehicles (AVs), games, factory automation (FA), decision support, navigation, medical diagnosis, and many others. Lastly, the rapid expansion of the “web” has meant that vast quantities of data from social media, sensors, and other sources can be routinely made available, often in real time, to train and test new ideas. Taken together, these represent a positive confluence of factors; AI is seemingly everywhere, being tried in almost every application, and the field is very dynamic.
There is an effort to get onboard, as AI is now widely viewed as a “strategic technology.” Governments, companies, and venture firms are investing in AI and rushing to create powerful new applications – with potential impacts for societal health and well-being, the economy, security, and many more. Of course, with new capabilities come new concerns about information ownership, security, privacy, trust, and the role of regulation by governments and vendors. Understanding what others are doing is essential, as evidence to inform policymakers and to help with business decisions.
This book provides unique and timely insights into AI activities spanning the Asia-Pacific region, focusing on Australia, China-mainland, India, Japan, Korea, Singapore, Taiwan, and Thailand. While our emphasis is on the breadth of research, we make special effort to include government policies, funding, workforce development, key institutions and projects, significant applications, large companies and start-ups, as well as challenges and strengths. There are many government-proclaimed national strategies for AI with “grand plans along with a vision of a great future.” These are described in appropriate detail. However, we have tried to provide a balance with reality, the actual strengths (or weaknesses) of the players on the ground, and the likelihood of realizing such visions.
There is a global race for AI leadership. Countries in Asia, including China, Japan, and others, are among the front runners. While there are significant differences across the region, there are also interesting commonalities, in particular a focus on “human-centric AI.” That is, an emphasis on AI “for society,” AI “with humans, for humans,” and more “human-like AI.” Also, many Asian players focus on real-world and practical implementations of AI, which strongly depend on local context, local environment, local expertise, and possibly “small data” rather than big data. In contrast, large U.S. players often focus on “Internet big data” AI.
All of us are familiar with Asia’s strengths providing the world with products incorporating the latest technologies. That is, being good at “bolting-on” a new technology and rapidly ramping up commercialization. However, since AI implies a more intimate human element, the introduction of products will require more nuance and care. Maybe toward addressing that issue, one priority that is common across the region is an emphasis on education and talent development for AI. In most other S&T fields, it is rare to see so many nations putting such a strong focus on education for a particular field - in this case, for AI-literate workforces. It is certainly a unique period.
Based upon extensive on-the-ground research, this book is an essential resource for corporate planners, government decision-makers, researchers, and anyone interested in how one of today’s most exciting new technology fields is developing in a dynamic part of the world.
