DECEMBER 6, 2019
1. The Formula
“THE THREE INVENTORS won the Nobel Prize for their efforts, but only one of them, Dr. Shockley, was determined to capitalize on the transistor commercially. In him lies the genesis of the San Francisco silicon story.” So begins Don C. Hoefler’s 1971 three-part series “Silicon Valley U.S.A.” for the trade journal Electronic News. With this series, the term “Silicon Valley” was born. Five years later, The New York Times shared the “Revolution in Silicon Valley” with the rest of world.
“Silicon Valley,” like many descriptions in the sciences (e.g., “the selfish gene”), serves not so much as a physical depiction as a metaphor — in this case, not as a label for Santa Clara County, but as the embodiment of a successful innovation hub. The term’s popularity has fueled efforts by academics and policy makers to unpack Silicon Valley’s secret sauce. In his history of venture capitalism, Harvard Business School professor Tom Nicholas recently distilled Silicon Valley’s success down to “the intersection of three main factors: direct and indirect benefits from universities, government military expenditures as a boost to high-tech, and a special cultural, legal, and physical climate.” A supportive media could be added to the list. While most experts recognize that Silicon Valley “grew out of a historically and geographically specific context,” as MIT researcher Timothy Sturgeon writes in “How Silicon Valley Came to Be,” that hasn’t stopped players from trying to emulate it.
Numerous US cities, for example, have tried rebranding themselves to attract Silicon Valley talent and funding. There is now a “Silicon Beach,” a “Silicon Desert,” a “Silicon Shire,” a “Silicon Praire,” and even a “Silicon Slopes.” Perhaps most successful is Seattle, which has been called Silicon Valley’s “little brother” or “Silicon Valley North.” Some commentators worry that it risks becoming Silicon Valley’s overflow zone. When Seattle-based Amazon recently sought a new secondary headquarters (HQ2), over two hundred US cities bid to host the company’s new home. Amazon declared it wanted to think “big and creatively” when considering locations and real estate options, and it identified two key geographic characteristics: “incentive programs available for the project at the state/province and local levels,” and proximity to a strong university system.
Hype around Silicon Valley is not a uniquely US phenomenon. Canada and Mexico were quick to bid on Amazon’s HQ2. Russia’s answer to Silicon Valley is the “Skolkovo project,” an “innovation city” on the outskirts of Moscow. When French President Emmanuel Macron took office, he told the world: “Entrepreneur is the new France.” He pledged to turn France into “a startup nation” and a “country of unicorns” — Silicon Valley slang for privately held companies valued at more than $1 billion. Israel was crowned the startup nation in 2009 for having the most startup companies on a per capita basis. But perhaps no nation has more Silicon Valley fever today than China. It even has plans for a “Greater Bay Area.”
2. A Star Is Born
Until recently, the path of digital innovation was viewed as having an eastward trajectory from the United States to Europe to Asia. While Western audiences have long recognized China’s rise — from proclamations of an “Asian Century” since at least the early 1900s to The Economist’s launch of a China section in 2012 (the first time the paper added a section in 70 years) — what is new is the growing visibility of Chinese tech giants such as Baidu, Alibaba, and Tencent. And new also is the recognition that these firms are legitimate competitors to their Western counterparts. Even Silicon Valley’s leaders, like Mark Zuckerberg, have become enamored by Chinese tech products, as Sue Halpern documents in The New Yorker: “If this [Facebook’s plan to merge all of its products] sounds familiar, it is. Zuckerberg’s concept borrows liberally from WeChat, the multiverse Chinese social-networking platform, popularly known as China’s ‘app for everything.’”
Along with the growing power of Chinese tech companies has come the same celebrity status for their entrepreneurs as for their counterparts in the United States. Chinese entrepreneurs now gloss the covers of Fortune and Time.
I met one of these celebrities at his book launch last fall in San Francisco’s Financial District. Surrounded by the San Francisco elite at the offices of the Asia Society, a global nonprofit organization with the mission of strengthening ties between Asia and the United States, Kai-Fu Lee was dressed smartly in a suit, and the belle of the ball. After a book signing for AI Superpowers, we gathered in an auditorium to hear Lee explain how new Silicon Valleys are emerging across the Pacific — ones that he predicts will overtake the United States’s, in particular regarding the implementation of artificial intelligence.
Lee is well positioned to make this claim. Born in Taiwan, he moved to the United States in 1972 at the age of 11 and attended Columbia University and then Carnegie Mellon, where he received a PhD in computer science. After a brief stint as an assistant professor at Carnegie Mellon, Lee left academia for Apple. As he writes in his autobiography Making a World of Difference, “I felt being stuck in an ivory tower, totally disconnected from the real world where I desired to make a difference.” In demand because of his research in speech recognition along with his bicultural background, Lee was able to work at Microsoft and Google in roles focused on expanding into China. Why US company expansions into the Chinese market have failed is key to AI Superpowers.
His insider knowledge has afforded him an immense following — both in China, where he has cultivated a “father”-like image as a mentor for Chinese students, and now in the United States, where Lee has been interviewed on CBS’s 60 Minutes and where AI Superpowers has become a New York Times best seller.
To make his credentials even more conspicuous, since leaving Google in 2009, Lee has directed a technology-oriented and Beijing-based venture capital fund, Sinovation Ventures, through which he has helped to create the future he predicts. In AI Superpowers, he details the three factors that he anticipates will lead China to AI supremacy and to hosting the next Silicon Valley. He claims that (1) elite expertise, particularly regarding AI, isn’t as important as it used to be, in part because scientific information is now global; (2) Chinese entrepreneurs face different and more cutthroat competition than their US counterparts; and (3) the Chinese government is backing this new innovation ecosystem in ways a capitalist country’s government could not. While Lee offers compelling evidence on each point, his book falls short when he couches his arguments in terms of a straightforward competition between the United States and China. He seizes upon the US-China schism, but fails to account for its origins. Nor does he account for how an international “decoupling” of the world’s two largest economies will affect their technological development.
3. An AI Revolution … Is Needed
Lee begins AI Superpowers with China’s “Sputnik Moment” — the moment AlphaGo, a product of the British AI startup DeepMind, acquired by Google in 2014, defeated top-ranked Go-player Ke Jie in May 2017. As Lee points out, “What you saw in this match depended on where you watched it from.” To the US business community, it was just one more confirmation that the West would continue to dominate technology innovation into the age of AI. But to Lee, and many in China, it was both a challenge and an inspiration.
AI became widely known to the Chinese business community in 2014, when Baidu, China’s “Google,” hired Andrew Ng from Google to work on the Baidu Brain, its artificial intelligence initiative. As a search term in the Baidu search engine, AI took off in early 2016. And less than two months after Jie resigned his final game to AlphaGo, the Chinese central government issued an ambitious plan to build artificial intelligence capabilities, building on its “Made in China” 2025 plan, which sparked the “trade war” with the United States.
The current enthusiasm for AI stems from deep learning — a technique that endows a machine with the ability to make educated guesses based on data. While the concept has been around since the 1950s, only in 2012 with the requisite hardware and data did deep learning become accurate enough to grow from a tinkerer’s curiosity to being intimately intertwined with the economy.
Lee argues that while deep learning’s birth took place almost entirely in the United States, Canada, and the United Kingdom, “China will be the biggest beneficiary of the heat the AI fire is generating.” He acknowledges that China is not yet as proficient as these countries when it comes to research. Even so, he contends that the coming changes in society will not occur from another fundamental breakthrough like deep learning. It will come from implementation — from harnessing the “low-hanging fruit” of big data. And no country has more data or more entrepreneurs attempting to implement deep learning than China.
But is that right? Lee acknowledges that we are only in the first of several waves of AI — one that seems focused on consumer convenience. AI will ultimately culminate, he writes, in autonomous AI, revolutionizing our “malls, restaurants, cities, factories, and fire departments.” Yet he believes that we can get there using current technologies. He is likely wrong.
Deep learning is a powerful tool. Its ability to learn through trial-and-error is at the heart of its success with Atari video games and Go. Beyond personalized ad, book, and movie recommendations, it accurately identifies and labels objects, recognizes speech, and issues loans. Its versatility is reflected in applications that range from optimizing energy usage in buildings to creating synthetic art in the style of Van Gogh. It will soon be able to assist physicians in diagnosing dieases and suggest treatments. But deep learning is not a trustworthy tool.
Deep learning systems begin as “blank slates.” They are only as good as the data we give them. And while Lee points out that no country collects more data than China, he conflates collecting with exploiting. Acknowledging and correcting unintended biases like race and ethnicity isn’t easy. Proxies abound while the system harbors an aura of objectivity. In addition, getting enough data is often impossible. In games like Go or chess, deep learning can reach super-human abilities because the rules are static. But in open systems (i.e., the real world), deep learning fails in inexplicable ways. While deep learning has markedly improved Google Translate, for example, it struggles because we create new sentences with fresh meanings all the time.
This is not to say that deep learning hasn’t produced a revolution. Lee is right that we have entered an “age of implementation.” But as AI researcher Pedro Domingos puts it, “People worry that computers will get too smart and take over the world, but the real problem is they’re too stupid and they’ve already taken over the world.”
Some experts contend that we are currently stuck at a “local maximum.” Gary Marcus and Ernest Davis write in their book Rebooting AI: “It’s not that it is impossible in principle to build a physical device that can drive in the snow or manage to be ethical, it’s that we can’t get there with big data alone.” In Artificial Intelligence, Melanie Mitchell echoes Marcus and Davis’s diagnosis and similarly cites the need for machines that can actually understand the situations they confront. Deep learning systems are often “good enough” in commercial settings such as for ad recommendations. But their lack of understanding reveals itself in the un-humanlike errors they make like, to use an example Mitchell offers, Google Translate rendering “I put the pig in the pen” into French as “Je mets le cochon dans le stylo” (mistranslating “pen” in the sense of a writing instrument).
To address this concern, the European Parliament has enacted a so-called “right to explanation” regulation — the right to an explanation for a given algorithmic output. Mitchell similarly calls for regulations focused on deep learning’s lack of reliability, transparency, and vulnerability to attack. These are essentially calls for another revolution — not to mention a societal reckoning of what kinds of explanations are feasible and meaningful — before we even begin to think of trusting AI with, for instance, driving cars.
Lee closes his discussion of this topic by stating that even if another AI revolution occurs, China still maintains the advantage because of the open nature of AI research. He is betting that the revolution will occur in the open space of academia rather than behind the closed doors of a Google, which would protect its trade secrets. This is an uncertain bet at best.
4. Copycat Nation
Lee maintains that China harbors two characteristics in particular that give it a long-term innovative edge over the United States: unparalleled competition among its entrepreneurs and government support.
One of these Chinese entrepreneurs, Xiaomi co-founder Lei Jun, told Wired magazine, “Don’t call me China’s Steve Jobs.” He may have imitated Steve Jobs when building the smartphone maker, but he now believes that he has created something new. And with the publication of Lee’s AI Superpowers, numerous outlets have reinforced Jun’s sentiment that something distinct is happening in China. But isn’t China the “copycat nation”?
Lee addresses this question head-on in a chapter called “Copycats in the Coliseum.” He admits that at the turn of the 21st century, Chinese entrepreneurs like Wang Xing were shamelessly copying America’s hottest startups. Wang, for example, copied Twitter, Groupon, and Facebook — going so far as to retain the tagline “A Mark Zuckerberg Production” at the bottom of each page. And yet, in 2017, when Groupon’s market cap was trading at one-fifth of its 2011 initial public offering, Wang’s “copy” Meituan had become the fourth most valuable startup in the world.
We would be mistaken, Lee writes, to assume that “Meituan triumphed by taking a great American idea and simply copying it in the sheltered Chinese internet.”
Wang didn’t build a $30 billion company by simply bringing the group-buying business model to China. Over five thousand companies did the exact same thing, including Groupon itself. The American company even gave itself a major leg up on local copycats by partnering with a leading Chinese internet portal [Tencent]. Between 2010 and 2013, Groupon and its local impersonators waged an all-out war for market share and customer loyalty, burning billions of dollars and stopping at nothing to slay the competition.
To Lee, copying, which would lead to social ostracism in Silicon Valley (although, to be sure, Silicon Valley is also fractured around intellectual property issues), was a necessary stepping-stone for Chinese entrepreneurs on their way to “re-innovation,” or creating more original and locally tailored tech products.
The transformation Lee describes is discussed more fully in Mark Greeven, George Yip, and Wei Wei’s Pioneers, Hidden Champions, Changemakers, and Underdogs. This volume claims that the current literature on Chinese innovation has left out much of today’s innovation landscape. Charting China’s entrepreneurs since the mid-20th century, Greeven et al. distinguish between four types of innovators as listed in their title, three of them commonly forgotten. While readers have likely heard of what the authors call “the Pioneers,” such as telecommunication equipment manufacturer Huawei and e-commerce platform Alibaba, they may be less familiar with, say, Weihua, an “underdog” that was founded in 2010 by three graduates of Tsinghua University. One of its founders received a PhD from Switzerland’s École Polytechnique Fédérale de Lausanne studying photovoltaic cell technology. In 2013, Weihua partnered with Merck, a German chemical giant, to develop a light, flexible solar cell that is more efficient than any other solar cell of its kind. Another example is Toutiao, a “Changemaker” founded in 2012 and now valued at $11 billion with 120 million daily active users. It uses artificial intelligence to provide mobile customized news recommendations and is disrupting state-controlled media.
Like Lee, Greeven et al. outline what can make Chinese innovation so disruptive. While Silicon Valley is reputed for competing on differentiation and strong technology, Chinese entrepreneurs follow the “swarm,” or, in Lee’s words, “competition in the coliseum.” In addition, Silicon Valley entrepreneurs often start with cutting-edge technology and then promote the product to market. Chinese entrepreneurs, on the other hand, are customer driven and listen to current market demands, often seizing on commercialized technologies and improving them. Think of the high-speed rail. (This means they skip research and development — an inefficient process to be sure but one that encourages resilience.)
Greeven et al. only briefly address the role of the state in creating these innovative companies. Lee, on the other hand, has a theory — and it doesn’t involve a protectionist environment. He claims that underestimating the “copycats in the coliseum” is precisely what led US companies to fail in China.
As evidence he writes that Lee’s Google China was outcompeted by Chinese search engine Baidu before the Chinese government blocked Google services in 2010. While Baidu’s core functions and minimalist design indeed mimicked Google, its founder Robin Li relentlessly optimized the site for the search habits of Chinese users.
Sans Silicon Valley companies in the Chinese market, China developed an alternate internet universe, according to Lee. After the iPhone’s 2007 debut, the tech world began adapting websites and services for access via smartphone. And when cheap smartphones hit the market, waves of Chinese citizens leapfrogged over personal computers entirely and went online for the first time via their phones. (Today, 98 percent of Chinese internet users are mobile internet users.) Lee writes that, for them, the internet isn’t an abstract collection of digital information that you access from a set location, but rather a tool that you bring with you as you move around. This new approach has led to the “O2O Revolution,” or “online-to-offline” revolution.
While Uber created one of the first transformational O2O models — ride-sharing — China expanded this model to dozens of other industries, including mobile payments and bike-sharing, a social and commercial transformation powered by Tencent’s WeChat, dubbed “a remote control for our lives.” As Lee writes, “Recent estimates have Chinese companies outstripping US competitors ten to one in quantity of food deliveries and fifty to one in spending on mobile payments. China’s e-commerce purchases are roughly double the US totals, and the gap is only growing.” In fact, one-fifth of the total world population online is now Chinese.
All this said, Lee’s description for why Silicon Valley hasn’t been able to expand into China is deeply unsatisfying. It doesn’t answer the question of whether Chinese entrepreneurs can be truly innovative. Can they only thrive in a sheltered ecosystem?
5. China Inc.
While working for Google, Lee used heat maps to track and better understand users’ online activity. He realized that American and Chinese users scroll through search results remarkably differently. American users spend about 10 seconds on each page before scrolling away, concentrating on the top few results, whereas Chinese users’ heat maps look like “a hot mess.” And this result makes sense. Americans treat search engines like the Yellow Pages, a tool for finding specific information. But the tens of millions of Chinese users new to the internet treat search engines like a shopping mall. As Lee writes, “That strikingly fundamental difference in user attitudes should have led to a number of product modifications for Chinese users.” But for any change to a core product, which would “fork the code” and make it more difficult to maintain, Google, like most Silicon Valley giants, has a lengthy review process. Therefore, instead of tailoring their products to Chinese users, American companies simply market their existing products.
Lee maintains that it is this slowness or inability to adapt to the Chinese market that explains the failure of most Silicon Valley giants to expand into China. But while Google, like any foreign company entering a market, faced a learning curve, Google faced other challenges as well. The Chinese government was not hands-off. In fact, Google operated in a highly regulated market and faced censorship, which led to China blocking access to YouTube in 2009 ahead of Google, and more generally in 2010 with the Great Firewall. The final straw, however, according to Google co-founder Sergey Brin, was a targeted cyber-attack, dubbed Operation Aurora, attempting to gain access to the Gmail accounts of Chinese human rights activists.
While the companies highlighted above may be successful in China, they mostly rely on second-tier innovations and face artificially limited competition. Richard P. Appelbaum, Cong Cao, Xueying Han, Rachel Parker, and Denis Simon write in Innovation in China, for example, that “WeChat’s large success is due, in some part, to the fact that almost all foreign social media platforms such as Facebook, Twitter, Instagram, WhatsApp, and Snapchat are banned.” Further, WeChat’s international expansions have been failures. Undoubtedly, one reason is that consumers’ usage patterns have already been set by products like WhatsApp in the United States and Blackberry Messenger in Indonesia. When I discussed with Lee whether Google’s latest foray back into China would be a success, he emphasized that Chinese consumers already have a suite of products that work. He then flipped the question on me: how different would Baidu have to be from Google for me to switch?
While a supportive policy environment is Lee’s final building block when it comes to creating an AI superpower, Appelbaum et al. question whether “the heavy-handed role of government” might, in fact, discourage innovation. As they highlight, behind the increased activity in research and development in China “lies the highly visible hand of the Chinese state.” In market economies, governments intervene to fund upstream research out of concern for market failure. As one scholar has put it, a government is a country’s “VC writ large” when it comes to funding basic research. But much of the government’s intervention in China is about innovation rather than basic research.
Appelbaum et al. write that the Chinese government is involved in “picking winners, prioritizing industries, and betting on manufacturing stars.” High-tech firms, for example, have to be certified by the government, which in turn grants them preferable policies. These firms, thus, do not innovate for the sake of development but instead to respond to the government’s policy incentives. The dangers of centrally planned economies are well known. Andrew McAfee recounts in his book More from Less the tragic effect of Stalin’s five-year plan on the whale population in the early 20th century: in calling for increased tonnage at fisheries, fisherman hunted whales in violation of international treaties purely for their weight. They had no other use for the whales.
China’s economic structure, however, is markedly different from Russia’s state capitalism. In a law review article, Harvard Law professor Mark Wu recounts how between 1997 and 2003 the state sold off nearly half of its state-owned enterprises (SOEs) because “Premier Zhu Rongji believed that the central government should focus only on supporting critical sectors, and that even in those sectors it should subject SOEs to market discipline.” Yet while private enterprise drives significant economic activity, the Party-state remains all-powerful. Private businesses with more than 50 employees are required to install a party secretary. Large companies such as Baidu even have members of China’s National People’s Congress in their ranks. China also continues to rely heavily on SOEs, particularly in the financial sector. Appelbaum et al. write that state-owned banks account for nearly three-quarters of all bank assets in China, and these banks in turn privilege state-owned enterprises over private firms. These SOEs “tend to be bureaucratic, risk (and hence innovation) averse, and beholden to party connections.”
These same issues extend into education and research. Take the so-called “brain drain.” While there are several pull factors (China’s rising economy, “talent programs,” increased opportunities for entrepreneurial ventures, and the rise of the sharing economy) and a number of push factors (effects of the 2008 financial crisis and increased difficulties in gaining permanent resident status in countries such as the United States), Appelbaum et al. write that creating a true performance- and merit-based culture will be the greatest pull for high-end returnees. In 2010, deans of life sciences at Peking and Tsinghua universities, Yi Rao and Yigong Shi, highlighted the problem in an editorial in the journal Science:
Although scientific merit may still be the key to the success of smaller research grants, such as those from China’s National Natural Science Foundation, it is much less relevant for the megaproject grants from various government funding agencies […] This top-down approach stifles innovation and makes clear to everyone that the connections with bureaucrats and powerful scientists are paramount, dictating the entire process of guideline preparation.
In addition, while China has seen explosive growth in scientific output — overtaking the United States as the world’s leading scientific publisher in 2009 — its publications and patents suffer in comparisions of quality. As Appelbaum et al. note, although China’s total patent filing growth dwarfs other countries, it remains far behind in terms of licensing revenue, patent filings abroad, and applications granted. Policies that incentivize the filing of patents without adequately emphasizing the importance of patent quality no doubt play a role. Some scholars have even noted a “Christmas rush” in patent applications when companies avail themselves of year-end government incentives for patent filings.
Lee ignores many of these challenges while highlighting the advantages of China’s state-led approach to science and technology. He is certainly right to highlight them. Given its techno-utilitarian approach, China can quickly create supportive policy environments for AI development without having to contend with the United States’s political gridlock. This advantage may play out with autonomous vehicles (AVs). While not as technologically sophisticated — Waymo’s cars alone have self-driven more miles than all Chinese AVs combined — Chinese firms may be able to sidestep certain issues in driving software because of AV-friendly urban landscapes. Lee cites Xiong’an as “poised to be the world’s first city built specifically to accommodate autonomous vehicles,” with potential adaptations like sensors in roads and traffic lights equipped with computer vision.
But Lee ignores the implications of such ease in implementing AI. While he approvingly compares AI to the harnessing of electricity and the lightbulb, AI should also be compared to a weapon. The greatest obstacle to China’s plan to create the next Silicon Valley is whether the international community will be willing to trust it with technological leadership. While foreign corporations used to worry mostly about business risks when entering China (e.g., intellectual property theft), today reputational risks matter almost as much.
6. Shaking a Great Nation
On the eve of the 80th anniversary of Silicon Valley’s original startup, Hewlett-Packard, “techlash” was crowned the 2018 word of the year. Some predicted that Silicon Valley would experience a “techsodus” in 2019. Others are now pondering whether Silicon Valley has lost its soul. Facebook, for example, was fined $5 billion in July 2019 for privacy violations. However, as The New York Times reported:
Levying a sizable fine on Facebook would go against the reputation of the United States of not restraining the power of big tech companies. For years, American regulators have faced criticism that they allowed Silicon Valley firms to grow unchecked, even as their European counterparts aggressively brought actions against tech companies — including fining Google a record $5.1 billion last year for abusing its power in the mobile phone market.
As this quote implies, lax regulations were key to Facebook’s early success. Now, those days are likely in the rear-view mirror. To understand the future of Silicon Valley, look to Europe, some commentators argue; it is pioneering a “distinct tech doctrine that aims to give individuals control over their own information and the profits from it, and to prise open tech firms to competition.” California seems to be first in line to follow.
President Xi would be wise to heed Silicon Valley’s moral panic. Not only is China currently the target of a trade war with the United States, but Silicon Valley entrepreneurs increasingly frown upon companies aligning themselves with China. Outraged Googlers, for example, forced Google to suspend project “Dragonfly,” a Chinese-censored version of the search engine. Silicon Valley, in other words, has been forced out of its historical amnesia.
While China may be collecting troves of data to power its AI research, it comes at the expense of serious privacy and security concerns. Human Rights Watch, for example, flagged iFlyTek, a Chinese AI company specializing in voice and speech recognition, for collaborating with Chinese authorities to pilot a surveillance system. Hikvision, a supplier of video surveillance products, has similarly been denounced for human rights concerns. Huawei itself continues to operate under a cloud of international suspicion that its telecom equipment includes a “back door” that is accessible to the Chinese government.
Huawei and other Chinese companies are extending olive branches in an effort to combat both the current trade war with the United States and accusations that Chinese companies are government surrogates. In an interview with The Economist, Huawei’s chief executive Ren Zhengfei, for example, has offered to sell the Chinese telecommunications giant’s 5G technology. The buyer could then modify the source code, eliminating any threat of Huawei or government control. Whether China would actually permit such a deal is another question.
President Xi isn’t only interested in displacing Silicon Valley. With his Belt and Road Initiative (BRI) global infrastructure project, Xi wants to reorganize the global economic order. His 2012 “Chinese dream” speech gave voice to China’s rising nationalism and its intention to rectify perceived injustices of the past. Yet Xi has sent mixed messages. A few months before his “Chinese dream” speech, Xi advanced a more cosmopolitan vision at the opening ceremony of the National Assembly of the International Astronomical Union. He declared:
Science and technology have no nationality! […] Nowadays the challenges for science and technology are more and more globalized, and all humankind is faced with the same problems in energy and resources, ecological environments, climate change, natural disasters, food security, public health, and so on.
Xi could have added AI to the list of coming problems. Lee, along with many other prognosticators, worry that as AI spreads across the global economy “leading to widespread technological unemployment,” the divide between the haves and have-nots will widen. Lee compellingly highlights the pressing nature of this problem. But, this said, to do so has by now become little more than a well-rehearsed cliché. More interesting coming from Lee would have been a candid discussion of the complexities behind the US-China AI arms race — the very skirmish that likely brought many readers to this book.
AI Superpowers demonstrates China’s great appetite for innovation. Stifling collaboration comes at tremendous cost. The blacklisting of China’s tech firms — most recently eight companies whose products the US Commerce Department claims facilitate surveillance in Xinjiang, including startups working on facial and voice recognition — spooks Western research partnerships. It fosters each nation’s desire for self-reliance, a “decoupling” that the international trade system was intended to combat. But tremendous costs also come from permitting China to push its vision of digital dictatorship. AI Superpowers feeds into the current popular discourse around US-China antagonism. Just as The New York Times’s 1976 article brought Hoefler’s insights to a broader audience, Lee’s work alerts us to China’s ardent desire to recreate Silicon Valley within its own borders — and to the possible consequences of that desire.
The author thanks Berkeley Law Professors Mark Cohen and Peter Menell for their encouragement and insights as well as the Berkeley Asia IP Project for its support.