After Chinese ride-hailing giant Didi drove Uber out of China in 2016, it threw support behind companies trying to do the same thing in other countries. Now, a global anti-Uber alliance of scrappy startups is leveraging Chinese money and expertise to thwart competition, while China’s influence has steadily grown in a range of markets. This Chinese strategy of backing “insurgent” artificial intelligence (AI) startups is a minor, but important, insight of Kai-Fu Lee’s 2018 book, AI Superpowers: China, Silicon Valley, and the New World Order. Lee calls the strategy localization, and predicts that it will reshape geopolitics in the AI era. It’s a compelling tactic to explore as U.S.-China competition draws increasing attention.
The globalization of artificial intelligence is picking up steam. By 2030, AI is expected to generate $13 trillion in global economic activity and boost cumulative gross domestic product by 16 percent. China and the United States are already driving much of this growth. Their homegrown AI companies have focused primarily on winning over domestic markets to date, but the vast majority of future users still live elsewhere, writes Lee. As these superpowers turn to global markets, they are adopting very different strategies. While Chinese AI companies tend to back startups that Lee calls local insurgents—the approach Didi took—the “juggernauts” of Silicon Valley are hoping their high-quality, but largely undifferentiated, products gain traction.
Lee argues that the one-size-fits-all approach that worked well in the internet age is less suited to the AI era. (The traffic patterns a self-driving car learns in Kansas probably don’t translate well to the streets of India.) There are important local distinctions in everything from public service delivery to financial services, two of the many fields AI is reshaping.
China and America’s distinct approaches to global markets reflect their distinct domestic political cultures. Lax intellectual property regulations in China make copycat competitors a constant threat and encourage Chinese companies to constantly innovate, even after winning significant market share. Lee describes in fascinating detail the ways in which successful urban Chinese entrepreneurs undercut each other to survive, steadily adapting products to changing consumer preferences. More than 5,000 different group-buying companies were vying for ascendency in China by 2011, the same year as Groupon’s initial public offering, Lee recalls. The ultimate victor among them, a company called Meituan, easily bested Groupon in China and ultimately conquered its Chinese rivals through a strategy of near-constant iteration.
The Chinese emphasis on deep localization is not unique to artificial intelligence. A recent South China Morning Post article explains how Chinese company Transsion, now the largest mobile phone supplier in Africa, captured a niche market by calibrating its phone cameras’ exposure for darker skin tones. As Chinese professor of management Changqi Wu explains in the article, “While in general China’s technology may still lag behind the US or other developed countries, it can maintain an edge in developing markets by serving them well.” Wu goes on to suggest that companies like Apple would probably avoid such an approach, owing to the high costs associated with localization.
Whether investing in local companies or customizing their own products for diverse markets, China’s more localized approach positions it to gain tremendous influence, reaping data and profits wherever its entrepreneurs and investors operate. Author Abishur Prakash wrote last month in Scientific American that Chinese healthcare company Ping An might eventually deploy its AI-powered “doctors” to Russia, for example, programming them to operate in Russia’s unique regulatory environment and creating an AI ecosystem Russia comes to depend on. In Lee’s view, Chinese companies will play a role virtually everywhere if China’s localized approach persists.
Such extensive Chinese influence is not inevitable, though. Lee thinks countries seeking to benefit from AI’s many applications might become like vassal states to the AI superpowers—an idea the Economist’s review of Lee’s book finds unlikely. The Chinese “no-holds-barred” approach is yet untested and may fall short of a more established rules-based system. The review also anticipates that more global users may develop ethical objections. Facial recognition technology that China employs to monitor ethnic minorities, for example, may face backlash even in developing democracies trying to close a large technological gap. Moreover, there remain concerns about the relative quality of Chinese AI. But the Chinese penchant for localization also suggests that Chinese AI companies might figure out how to fix privacy and quality issues as their iterative approach reveals that these things matter to many consumers.
Artificial intelligence will soon permeate remote corners of the developed and developing worlds. Chinese know-how and funding will power much of it—though just how much remains to be seen. New technology will often improve quality of life and industry competitiveness, but inequality within and between countries will certainly grow as entire industries are restructured. Moreover, unique strategic advantages will accrue disproportionately to the owners of the most powerful algorithms. Non-superpowers should proactively establish a national AI strategy that defines their own comparative advantage in what scholars writing for the Carnegie Endowment have dubbed the “machine learning value chain.” Establishing how data moves between countries and addressing privacy concerns should be on every national agenda. Catching up to the AI superpowers on every front may be unfeasible, but preparing for massive changes ahead—and the tactics that will be employed along the way—is well within any state’s reach.