by Emma Zhang
Kai-Fu Lee, AI Super-Powers: China, Silicon Valley, and the New World Order, Houghton Mifflin Harcourt, 2018. 272 pgs.
In his 2018 book AI Superpowers: China, Silicon Valley, and the New World Order, Kai-Fu Lee, the former president of Google China, and current Chairman and CEO of Sinovation Ventures, provides a glowing review of the rapid development and adoption of the latest AI technology in China and discusses what it means for Silicon Valley as well as the world. Born in Taiwan and educated in the United States, Lee devoted most of his professional life to the advancement of AI technology and is therefore understandably keen on promoting AI-friendly policies and AI-related business ventures worldwide. He argues China’s unfailing national support for AI technology championed by the central government can possibly turn Zhongguancun, Beijing’s high-tech centre, into the next AI powerhouse comparable to Silicon Valley, but with distinct Chinese advantages that the Valley may not be able to compete with. Despite his optimistic overall appraisal of the power of AI to transform our lives for the better, Lee also earnestly alerts readers of the looming threat of AI replacing human labour and disrupting current socio-economic and political structures.
In 21 Lessons for the 21st Century (published August 2018, merely one month before AI Superpowers), the world-renowned Israeli historian Yuval Noah Harari cautions readers of an AI techno dystopia. Harari declares that the onslaught of AI replacing human labour will lead to the creation of a massive “useless class”—people who can never generate enough economic value to support their own lives. Since such a mass scale disempowerment of people has never fared well in history, Harari warns us that the threat of AI has become one of the most urgent concerns of our age. Lee, as a leading AI expert based in Beijing, confirms Harari’s dismal vision and makes the threat all the more concrete by examining the four waves of AI development, and listing which sectors of human labour will be replaced in what fashion. Unlike Harari, who argues for the continuous training and re-training of human workers so that they can better cope with constantly shifting labour markets as one of the solutions to the AI threat, Dr. Lee points out that this approach, though necessary, is insufficient, as it puts workers “in a state of constant retreat, like animals fleeing relentlessly rising flood waters, anxiously hopping from one rock to another in search of higher ground.” Instead, he offers an unexpected solution based on a deeply personal experience in confronting his own mortality.
Lee’s attitude towards AI development and adaptation is clearly informed by utilitarian pragmatism, which is utterly unconventional compared with other commentators. As such, what is typically frowned upon by others is praised by Lee. For example, in Political Order and Political Decay (2014), political scientist and historian Francis Fukuyama states that one of the factors that could impede the development of China in the age of InfoTech is its lack of freedom and corresponding inability to nurture true innovation. Lee presents a completely opposite view. He argues that the innovation and adaptation of AI technology is actually hampered by intellectual property laws and the system of democracy. He points out that Silicon Valley’s starry-eyed mission-driven philosopher-inventors regard copying “as a betrayal against the zeitgeist and an act that is beneath the moral code of a true entrepreneur”; whereas in China, copying is the normal practice of the battle-hardened profit-driven gladiatorial technician-businessmen. This means that although Chinese technicians may lag in their ability to truly innovate, their capacity to mimic existing AI technology—to the degree that even highly trained experts have trouble telling the copy from the original—is nonetheless of great value.
According to Lee, the benefit of innovative ability is too often exaggerated, and in truth, innovation provides limited advantage—especially in a market where not all the players follow the same rules. Game changing innovations come rarely, once every few decades at best. In AI tech, the latest breakthrough was deep learning, and a plateau has been reached since then. Lee calls the current phase of AI development the age of implementation. In this stage, the groundbreaking Nobel winners may not produce as much advantages as large teams of well-trained AI scientists Lee calls “tinkerers” who adopt AI into everyday use, such as loan assessments, self-driving cars, computer aided translation and product promotion. In addition, thanks to the openness of the AI research community that made new discoveries instantaneously available worldwide, entrepreneurs in China can safely claim that China lags behind Silicon Valley in AI research by a mere sixteen hours—the time difference between California and Beijing. In this environment, the country that adopts AI with the greatest effectiveness will harvest the most value. Lee therefore reasons that China’s AI friendly policy, cutthroat market driven competition and growing number of users generating enormous data will no doubt give the country a distinct advantage. China, consequently, is on its way to becoming one of the world’s most formidable AI superpowers.
Furthermore, in Chapter 4, “A Tale of Two Countries,” Lee compares government support for AI in China and the United States and argues that China’s centralised authoritarian power structure can in fact create a virtuous cycle for effective AI deployment. He provides the example of a monotonous speech given by Chinese Premier Li Keqiang in 2014, which generated “a raging fire in the Chinese technology industry, pushing activities in the investment and startup space to feverish new heights.” In contrast, Obama’s 2016 AI implementation plan press release was overshadowed by Trump’s infamous Access Hollywood videotape and consequently “barely registered in the American news cycle.” Worse, after Trump’s election, which was merely three months after Obama’s launching of the AI plan, president Trump proposed cutting funding for AI research. This shows that America’s restricted governmental power and shifting leadership can negatively impact the AI industry and restrict its development; whereas China’s centralised power structure can more effectively mobilise the entire nation’s resources, both human and material, into promoting rapid AI development and adaptation.
Another rather disturbing example of China’s authoritarian political structure giving rise to more rapid AI acceleration is demonstrated by Lee with the case of self-driving cars. In Industries of the Future (2016), the American technology expert Alec Ross suggests that self-driving cars will not be adopted (in America) anytime soon, because though the technology reliably reduces traffic accidents caused by human error, people will be more willing to accept accidents from a flawed person rather than a smaller percentage of errors caused by a “black box you cannot control.” Dr. Lee, however, claims this technology is likely to become widely available (in China) in “the not-too-distant future,” because the Chinese government will not “let the perfect be the enemy of the good.” The authoritarian Chinese system means there is only upward accountability. Thus, as long as the Chinese government considers the implementation of self-driving technology worthwhile, the corresponding massive job loss for drivers, and the relevant ethical concerns (such as whether the vehicle should be programmed to save its own driver or save a larger number of pedestrians) can be worked out along the way. Lee concludes that China’s higher acceptance for risk will put the country ahead in the AI race.
Furthermore, China’s authoritarian hold on power also means the government has greater liberty in allocating resources that no democratic countries could match. In Chapter 5, “The Four Waves of AI,” Lee states Zhejiang province is planning to build “the country’s first intelligent superhighway, infrastructure outfitted from the start for autonomous and electric vehicles.” Beating the superhighway is “Xiong’an New Area” sixty miles south of Beijing, which is an area the size of Chicago destined to become “a showcase city for technological progress and environmental sustainability” as well as “the world’s first city built specifically to accommodate autonomous vehicles.” Lee demonstrates that China’s lack of democracy means the government has greater ability to disregard the potentially disastrous risks caused by self-driving cars. Moreover, its exclusive control over resources, including both land and money, allows the country to build the infrastructure that accommodates this technology.
If Lee is correct in his judgements, then this book not only makes a case for AI-friendly policies but also (perhaps unwittingly?) puts the system of democracy in a degree of danger. Authoritarianism is frequently presented as being more efficient and effective whereas democracy tends to create stumbling blocks for the implementation of AI. Fortunately, the book is more of an earnest expression of Lee’s personal opinions rather than accurate prophecy of the future. For instance, in Chapter 3, “China’s Alternate Internet Universe,” Lee praises the bike sharing startups that must have been blooming when he wrote the book. However, today, many of these companies (including ofo) have since ceased services around the world, leaving a massive number of unclaimed bikes littered all over the cities in which the formerly operated. Unfettered power can generate seemingly efficient adaptation of a new technology, but such “raging fire” can also lead to massive failures in implementing immature systems, causing colossal waste. The patient caution that democratic systems assert may lead to better designed and more sustainable use of new technology in the long run.
Lee’s eager promotion of AI technology also makes the book seem blatantly lacking in discussion of ethical concerns surrounding emerging technical capabilities. For example, when illustrating the possible deployment of facial recognition (perception AI), Lee describes a pleasant shopping experience set in a future shopping mall where AI-powered shopping carts greet customers by name and sort out their shopping list based on their individual consumer habits. There is no contemplation about how such a technology could also empower a super surveillance state, and no discussion on what might happen when such technology is deployed by an authoritarian government like China with no checks or balances on its power. Furthermore, the shopping cart in Lee’s imagination is programmed to speak with the voice of Jennifer Lawrence—his favourite actress, just to make the chore more pleasant. There is no concern about how Jennifer may feel about her voice being used in this way—this example merely puts a female Hollywood star in servitude without consent, and though one could argue it is fairly harmless, anyone with the least bit of imagination can envision celebrities and public figures’ voices and images being abused in far worse ways.
The most valuable contribution of this book is Chapter 6 “Utopia, Dystopia, and The Real AI Crisis,” which consists of Lee’s careful analysis of how AI will affect the job market. He reports on the various estimations made by other experts, evaluates their relative credibility and provides convincing analysis of his own. Lee confirms with far greater certainty many concerns that Harari raised in abstract terms—AI will bring about unprecedented productivity and generate a world of plenty, but human beings will face massive scale job loss that can potentially turn current social-economic and political structures upside-down. AI will create winners and losers. Economic disparity will dramatically widen both within a country and between different nations. But even in this part of the book, Lee is overwhelmingly positive about China’s prospects. He argues that AI will likely replace white-collar jobs but poses less of a threat to blue-collar jobs. Although deep learning enables algorithms to make accurate calculations far beyond human capability, robotics has a long way to go before machines can mimic the intricacy of human movements. For this reason, Lee asserts, AI threatens the American job market more than it does China because about 50% of China’s labour market consists of blue-collar manual work. However, how will China’s white-collar labour force cope with the looming threat of AI? When these jobs are gone, what will happen to the millions of university graduates who already have trouble finding suitable positions today? And if AI cannot adequately replace blue-collar jobs, how will China survive its oncoming population crisis? The benefit of AI for China may not be as great as Lee envisions.
The final three chapters of the book are the most intimate and most moving, in which Lee describes his own encounter with cancer and how facing death has brought him a revelation that can possibly redeem the human race from a AI generated job loss crisis. He relates that being a cancer survivor has taught him that life is not about chasing after career advancement but about building meaningful human relationships. He says the ability to share love is what makes us human. AI will create a world of plenty, and human beings will, for the first time in history, be freed from the toil of Adam, and be able to live more fully by devoting more love to those around us. Lee imagines a kind of return to Eden, where AI works side by side with humans, each doing what they are created for—in AI the accuracy and precision, in humans the loving kindness. In the 18th century, amidst the first industrial revolution, where humans and machines were competing for jobs for the first time, rationalist philosophers were questioning whether human beings were mere machines programmed to maximise their own chances of survival. In response, Goethe wrote The Sorrows of Young Werther (1774) to remind his readers that human beings are far more complicated creatures than machines because of their capability to love. Today, Lee is reminding us of the same lesson. Will our need for power be conquered by our capability to love in the age of AI? I sincerely hope we can learn the lesson of love before the world whirls into a super surveillance Dystopia, where the AI powered Übermensch deem every other human being dispensable.
Emma Zhang is a lecturer of English in the College of International Education in Hong Kong Baptist University. Her research interests include comparative literature and comparative mythology. Her doctoral dissertation “Domination, Alienation and Freedom in Ha Jin’s Novels” (2015), analyses Ha Jin’s novels in connection with contemporary Chinese society. Her other works include “Father’s Journey into Night” (2013), “No End in Sight – the myth of Nezha and the ultra-stable authoritarian political order in China” (2018), and “The Taming of the White Snake – The oppression of female sexuality in the Legend of the White Snake”. She is currently working on translating ancient Chinese legends Nezha and The Legend of the White Snake.