By Geoffrey Cain
Mekong Review
May 2019
KAI-FU LEE AI Superpowers: China, Silicon Valley, and the New World Order Houghton Mifflin Harcourt: 2018
Ke Jie was slumped in his seat in May 2017, rubbing his temples, sighing and nervous as he pondered how he’d place his next stone on the nineteen-by-nineteen lined board. The game, called Go, is bewilderingly complex, and Ke Jie, from China, was its global star. “No human on earth could do this better than Ke Jie,” Kai-fu Lee, the author of AI Superpowers: China, Silicon Valley, and the New World Order, writes, “but today he was pitted against a Go player on a level no one had seen before.”
After three gruelling days and on his third and final game, Ke Jie capitulated. But in conceding defeat, he couldn’t shake his opponent’s hand. That’s because his opponent was an artificial intelligence program called AlphaGo, developed by a start-up called DeepMind that was acquired by Google. Artificial intelligence had done something once thought impossible: outwitted a champion — and all without feeling, consciousness or an understanding of its own significance.
The match marked a turning point for our relationship with technology and where it’s taking us, Lee argues. For years, AI engineers had regarded a software program that could win at Go as hopelessly complex: the number of possible positions on the board exceeds the number of atoms in the visible universe. The human brain was able to feel out moves and strategies, an intuition that a machine lacked.
China responded to Google’s dethroning of its champion with a Cold War-style technological race against US might. But the match went pretty much unnoticed in the West.
Lee, a prominent and respected Taiwanese venture capitalist in Beijing, takes us on a gripping journey into China’s ambitious start-up scene, where AI visionaries and tinkerers are mounting their challenge to Silicon Valley. Lee has the ideal background to write this book. He’s gone deep into both China and America’s technology cultures as a computer scientist. After earning his PhD from Carnegie Mellon University and becoming a leading researcher in voice recognition technology, he ascended through Apple’s Cupertino headquarters in the early 1990s and later Microsoft. By 2009, he had turned his sights to China’s nascent start-ups, setting up his venture capital firm, now called Sinovation Ventures, and overseeing a $2 billion portfolio.
For Lee, the arrival of the “AI revolution” is due to a mishmash of advances. Powerful semiconductor technology has elevated computing power, now able to crunch unprecedented amounts of data from social media and mobile apps. Software developers, meanwhile, have seized on a revolution called deep machine learning, a programming method that teaches AI to learn by example, closer to how the human brain works, rather than by applying a rigid set of rules
While Silicon Valley has a robust technological ecosystem, Lee believes that China is poised to catch up. First, he reasons that China’s “gladiatorial entrepreneurs” have come of age in a bloody, no-holds-barred cage fight, toughening them up against larger Silicon Valley rivals. He plots the rise of China’s “unique internet ecosystem” from its citizens’ disinterest in online privacy and their enthusiastic embrace of mobile payments. Both characteristics allow, on a far greater scale than in the US, the gathering of data that can be fed into AI software that can locate patterns in an endless deluge of data. Finally, he argues, China has a leg up because its government is “proactive” in state planning. It looks further into the future and pushes far more aggressively than the White House.
The result is that some of China’s AI pilot projects are clearly out-innovating those in the US. Video conference cameras in elementary school classrooms can see whether children are bored or listening, and the AI can adjust homework assignments based on whether they’re struggling. Court judges are participating in a pilot program in Shanghai in which AI software advises them on how to sentence criminals. The idea is that the software will act as a check on the judges’ unconscious biases, such as whether a suspect on trial is well dressed or poor, or comes from a certain ethnic group.
It was February 2018. I was driving through the barren, rolling frontier called Xinjiang with my guides. We were on our way to meet Chen Quanguo, the Communist Party hardliner in his sixteenth month as Xinjiang’s party boss. I was curious to see first-hand his experiment in how to control people using artificial intelligence. As we travelled through police checkpoint after checkpoint, we were frequently stopped, and police asked questions about my beard — beards were banned for younger residents as a measure against the Muslim population — and religious preferences.
I asked one of my guides how the Chinese state set up such an “admirable counter-terrorism effort”
“It is quite simple,” he said. “We have a pro-active government that takes terrorism seriously. We have very strong technologies and strong companies that help the nation.”
“Do the Chinese care about privacy on the internet?” I persisted.
“China is unique. We do not care about privacy like Americans. It’s a situation we accept for our safety in Xinjiang.”
China’s President Xi Jinping declared, “Advanced technology is the sharp weapon of the modern state”. In AI Superpowers, Lee’s much trumpeted “gladiatorial entrepreneurs”, “pro-active government” and “unique internet ecosystem” of mass data gathering ring like this sort of state propaganda. These three forces are building the world’s vastest AI-powered surveillance state in the far western frontier of Xinjiang, where 10 per cent of China’s Uyghur population, or 1 million people, languish in thousands of high-tech concentration camps.
What’s troubling is that Lee and other Chinese entrepreneurs brush aside questions on AI’s role in China’s ascendant authoritarianism. His book makes no mention of China’s AI-assisted “predictive policing” program, which allows for the arrest and detention of people who supposedly will commit a crime, based on demographics such as ethnic group and religious affiliation. He doesn’t mention the mass, forced data gathering on Uyghur Muslims or the AI-driven voice recognition database being developed that can identify anyone from their phone conversations
When the US current affairs program 60 Minutes asked Lee in January whether governments might deploy AI to control populations and crush dissent, he said: “As a venture capitalist, we don’t invest in this area, and we’re not studying deeply this particular problem”.
To overlook the scenarios in which the technologies you praise could be turned against humanity is troubling. In 2017, China passed the National Intelligence Law, requiring Chinese organisations — Lee’s Sinovation Ventures would be no exception — to “support, assist and cooperate with state intelligence work according to law”. If China’s AI developers are unwittingly creeping towards a dystopia of total control, they won’t be celebrating a landmark triumph that Lee likens to the space race. Rather, they’ll be floating in the dark depths of their true purpose.
To his credit, Lee urges us to prepare for another impending cataclysm, the widespread unemployment of both white-collar and blue-collar workers in manufacturing, law, management and even radiology, and the resulting social instability. “Within ten to twenty years,” he writes, “I estimate we will be technically capable of automating 40 to 50 per cent of jobs in the United States.” Poorer countries in Southeast Asia and South America that depend on cheap labour will fare even worse, since factories staffed by robots would be relocated to the US and Europe, closer to consumers and therefore cheaper.
Perhaps the answer is a universal basic income, paid for with a heavy tax on artificial intelligence companies. But Lee dismisses that. As he points out of the Silicon Valley chief executive officers who love the idea, “People who have spent their careers preaching the gospel of disruption appear to have suddenly woken up to the fact that when you disrupt an industry, you also disrupt and displace real human beings within it.” Universal basic income means an elite caste will be sedating the underclass with money akin to opium. It doesn’t address the fact that it’s still a caste society.
In one of his book’s stronger threads, Lee seeks an answer by telling the story of his encounter with his own mortality, being diagnosed with stage four cancer. After a lifetime of thinking like a machine, he realised that an inner force of empathy and humanity separates us from machines, no matter how smart they become. “I told myself that however much time I have left … I wouldn’t live by internal algorithms or seek to optimize variables. I would try to share love with those who had given so much of it to me, not because it achieved a certain goal but just because it felt good and true.”
As machines take over the human tasks of calculation and optimisation, our minds and efforts will be freed to embrace the things that make us human in ways never seen. Doctors can’t compete with machines in their ability to optimise treatment regimens, but they can take the role of what Lee calls “compassionate caretakers”, comforting and uplifting lonely patients. Teachers can’t compete with the speed at which AI can grade exam scores and measure progress, but they can spend more time mentoring each student, pushing the next generation towards a brighter future.
The answer could be what Lee calls a “social investment stipend”: a decent, guaranteed salary for people who invest their time into promoting a “kind, compassionate, and creative society”. To some, his views may sound lofty and far in the future. But it’s one of the more creative suggestions for a disruption that could be coming sooner than we think.
Geoffrey Cain, a technology writer, is a contributing editor of Mekong Review
The article was originally published in Mekong Review