Arushi Singh, Research Intern, ICS
From his perch at Zhongguancun, Beijing’s Silicon Valley, Kai-Fu Lee, the author of AI Superpowers: China, Silicon Valley, and the New World Order, delves into various nuances to explore the development of artificial intelligence (AI) prowess of Google’s AlphaGo that also showcases the versatility of the intuitive pattern recognition technology which smoothly segues into the incoming AI wave. The Chinese government is not far behind in taking the helm of all AI affairs. It has launched various ambitious programs also on “clear benchmarks for progress” for AI development. This response, in part, has been triggered by the swift acceleration of real-world AI applications in areas such as speech recognition, vaccine development, and applied natural language processing, which has rendered the book’s subject matter immensely consequential.
Lee’s wealth of information originates from his research conducted under the guidance of AI maverick Raj Reddy, to whom he has dedicated this book. In the 1980s, Lee programmed the first AI gaming software to defeat a member of a world championship team (Othello, a simplified version of Go) and Sphinx, the very first speaker-independent program. He illustrates how modern advanced AI systems are powered by deep learning that has “turbocharged the cognitive capabilities of machines”. The author further elucidates that AI winters commenced due to a lack of data that inadvertently resulted in reduced funding. However, technological advancements have solved the problem of paucity of data to a large extent.
AI Evolution at a Glance
The author delves deeper into the evolution of two AI approaches, i.e., the “rule-based” approach that focused on the encoding of logical steps and the “neural networks” approach that emphasised the construction of “layers of artificial neurons that can receive and transmit information in a structure akin” to humans’ neural networks. Lee also aims to dispel rumours regarding the prodigious development of AI in the recent decades; instead, the author highlights the innovative application of decades worth of research that focuses on the buttressing of learning and transfer learning that comprises and have been instrumental in constructing the current misguided AI perceptions.
Invoking Robert Mercer’s phrase that “there’s no data like more data”, the book emphasises the treatment of data that fuels AI. Mercer has formerly worked as an IBM language recognition specialist and later, the Co-CEO of Renaissance Technologies, a quantitative hedge fund firm. Notably, this emphasis on data in the book was exemplified by the author focusing on data’s role in pattern recognition and outcome optimisation through “narrow AI” powered by data.
The book has eight succinct chapters, organised into distinct themes that are the culmination of Lee leveraging his extensive AI background to secure the opportunity to establish Sinovation Ventures and invest in multiple companies in China. The first part is on the evolution of AI development in China. The second is the differences in AI development approaches in the US and China, from a private market and governmental perspective. The third section assesses the gradual development progression of AI or “AI waves”. The last portion of the book contends with integrating AI and humans to empower humanity. Lee shows prescience regarding AI development in China. For instance, Baidu has been making great strides in its AI ventures since the book was published, such as its LinearFold AI algorithm, Apollo Go Robotaxi service, ERNIE-GEN, Paddle Quantum, and Quantum Leaf.
AI Development Practices in US and China
The author’s personal experience shed light on the cultural affinities, the socio-economic atmosphere, academic norms, and government regulations propelling Chinese AI development, particularly in his aptly named “Copycats in the Coliseum” chapter. He differentiates between the American and Chinese practices that mean a world of difference in incubating future “AI giant”. These additional and ingeniously developments showcase the remarkable commitment and advancements in AI research that are ready to stand shoulder to shoulder with US AI research efforts.
Lee attempts to dispel the myth that China is still stuck in its “pixel-for-pixel” copying phrase and states that “pure copycats never made for great companies”. As stated by Lee, the reality is more complex, and Chinese companies have begun to move beyond mere copying. They have had to innovate in a highly competitive environment at a face pace, led by “gladiator entrepreneurs” who are intricately involved in innovation and reiteration.
Furthermore, one of the most important discussions is the technological culture difference between China and the US. Silicon Valley, to Lee being “downright sluggish” compared to its Chinese counterpart. Paradoxically, to stay in the AI race, constant innovation to achieve every product iteration necessitates a culture of continuous copying in China. Additionally, Lee also points out the strategies employed by giants in gaining their market share and how these strategies could outperform companies in the AI race. However, many such strategies and tactics are anathemas, especially in the US.
Overall, the book provides a comprehensive understanding and helpful overview of the existing realities of China’s AI research. The author has attempted to prescribe ways to deploy AI more effectively in the future deftly. The book’s last section gives the reader the factual assessment of the “AI race” brewing between the US and China. The book was woven together by anecdotes from Lee’s time as a venture capitalist and by his invaluable experiences as part of various companies such as Apple, Microsoft, Google and Silicon Graphics International during the course of his career. Therefore, Lee has been able to render a more vivid picture of the AI landscape in China and the US.
This book attempts to provide a balanced evaluation of the incremental gains made by companies in AI both in China and in the US, along with the global repercussions of their consequent “game-changing AI products”. Moreover, all the chapters display a firm grip over identifying the key drivers of AI development and investment. However, there is a greater focus on the personal trials and tribulations of the author that drives the focus away from the larger analytical framework focused on the topics of AI. While informative, the book’s limitation surfaces in its absence of inclusion of an in-depth analysis of the US government’s AI policy implementations. Likewise, the author does not care to discuss some of the significant AI hubs of China in provincial cities such as Chongqing, Chengdu, and Guiyang, some of the regional towns.
Nevertheless, the work is highly pertinent, and such niche themes have seldom been explored realistically. He suggests that AI-powered technologies could harness universal basic incomes by incentivizing “socially beneficially activities”. The book’s strength lies in its unique exploration of China’s “data-scape” in granular detail. It lays the foundations for further scholarship on the topic, which are pivotal as humanity is marching towards what Lee describes as “the quantification of the human thinking process, the explication of human behavior”, a subject that becomes more relevant by each passing day.