ounter-intuitively, AI Superpowers is not just an AI book. Nor is it a book on geopolitics. It does much more than critically analyse and compare AI developments in China and the United States. Far from perpetuating dusty stereotypes of China as an intellectual property thief, Kai-Fu Lee channels his experience as the founding director of Microsoft Research Asia and the president of Google China to provide an insider’s account on China’s technological rise. His account is a rich, honest, novel insight — and even defence — of the Chinese model and its impressive nurturing of a billion-dollar tech industry that characterises the image of modern China.
Lee structures his argument to first contextualise how China is where it is; able to propel ahead in global AI dominance. Most importantly, Lee charts the specific political, cultural and economic factors that have moulded China into a hub of hot data. The abundance of business insights buried within the chapters of AI Superpowers adds to its quality: for instance, through the government’s national zoning laws and investment policies, tech has been embraced rather than lambasted. To put it simply, the Chinese worldview reflects the inclusive sentiment of 'The People and Tech', whereas the American worldview increasingly treats the sector as an object of containment, 'The People v Tech'.
The American-Chinese techno-cultural relationship is a complicated one. In some respects, Lee argues that Chinese entrepreneurs have always emulated — and imitated — the culture and products of Silicon Valley. He terms the Chinese equivalents of American firms as “copycat” companies: Baidu imitates Google; RenRen copies Facebook; Meituan takes after Groupon; Toutiao is the copycat of BuzzFeed, Tujia after Airbnb, Didi Chuxing the equivalent of Uber, and Alibaba is the online marketplace behemoth inspired by Amazon. Where the U.S. has 'FAANG' (Facebook, Apple, Amazon, Netflix, Google), China has 'BAT' (Baidu, Alibaba, Tencent).
But in many other respects, Chinese entrepreneurs now surpass their American counterparts. From Tencent’s all-inclusive super app WeChat (described as a “digital Swiss knife”) to electronics company Xiaomi, Chinese tech entrepreneurs do the one thing that Silicon Valley overlooks: product iteration. Beijing’s ‘Zhongguancun’ goes further than Silicon Valley through its culture of constantly iterating a product or service to suit the customer. They remain in the business of services, rather than the business of platform-creation: even if that means dabbling with inventory and owning stock. Short video app TikTok is another illustration of Chinese iteration, its success is largely due to the wealth of recording, remixing and dueting features it offers, in part due to its earlier acquisition of lip-synching music platform Musical.ly. Further, instead of catering to the vast and divergent needs of the international consumer, China has leveraged the size and scale of its domestic market — a force strong enough to banish international behemoths Uber and Google as uncompetitive and unneeded aliens.
Lee then dives into the main substantive: the present and future applications of AI. The main ingredients of AI are: an abundance of data, a strong algorithm, a narrow domain and a concrete objective. The country’s revolutionary leap from cold hard cash to mobile payments is one example of how the country emulated the culture of Big Data — and thereby satisfied the first component of the AI formula. By restructuring its society to maximise data collection, China has built a national infrastructure that is designed to leverage AI.
Rather than referring to AI as a homogenous term, Lee's taxonomy builds on four categories: internet AI; business AI; perception AI (online-merge-offline services) and autonomous AI. His main substantive argues two main points. One, Chinese dominance in perception AI will grow to outweigh that of the United States’, although both will equate each other in autonomous AI. Two, the best hope we have for coexisting with AI is our social and sentimental edge, rather than any other aspect of analytical intelligence. This is because algorithmic ability to spot deep correlations hidden in Big Data means it is both redundant and zero-sum for humans to compete on this: eventually even sacred acts like medical diagnosis will be competently dealt by autonomous AI.
Yet far from the apocalyptic tone too many AI books exude, Lee possesses the rare skill of discussing futuristic events and issues with balance, sophisticated nuance and a clear-sighted ability to solve problems. He explains how deep learning works and flirts with the idea of the 'singularity' or artificial general intelligence (AGI): an eventual state where a computer fully and finally surpasses human capabilities. In spite of this, Kai-Fu Lee's techno-optimism proves contagious and, backed by vast amounts of research, desperately warranted.