A Teenager's Drive to Beat a Computer Program for Domination Over Artificial Intelligence in China

FE_TrumpChina_08_1079012834
Digital Human Brain Covered with Networks Andriy Onufriyenko/Getty

Lee Kai-fu, the former president of Google China and now a prominent venture capitalist in Beijing, remembers the scene well. Sitting in his office on a spring day two years ago, in an area of town known as the Silicon Valley of China, he, like millions of other people throughout the country, was glued to the television. On the screen, a gawky teenager in square black-rimmed glasses was squirming uncomfortably.

The young man was Ke Jie, and, as Lee jokes, he seemed an unlikely hero “to make humanity’s last stand.” But there the 19-year-old was, playing a game of the ancient Chinese board game Go. His opponent? A computer named AlphaGo, powered by an artificial intelligence program from a British startup, Deep Mind, that Google had acquired.

Go is believed to be the oldest board game, invented more than 2,500 years ago. Played on a 19-by-19-inch square board, it involves two players, one with white stones and the other with black, who try to encircle each other’s pieces. Fans call the game far more sophisticated than chess, imbuing its players with “a Zen-like intellectual refinement and wisdom,” as Lee puts it. 

FE_TrumpChina_09_631790790 Go player Ke Jie arrives at red carpet of Sports Personality Of The Year 2016 awards ceremony on January 15, 2017 in Beijing, China. VCG/Getty

Ke was a prodigy: He had started playing professionally at 10 years old, and by the time of the AlphaGo match, he was widely considered the best in the world. But in the Google event, with most of China watching, he was getting demolished by a machine, losing three straight games.

It was, as Lee describes it, China’s Sputnik moment. The reference is to the 1957 satellite launch by the Soviet Union, which stunned Americans. Their Cold War adversary had done something technologically that the U.S. was still attempting. It led to the creation of the National Aeronautics and Space Administration and President John F. Kennedy’s pledge that the U.S. would put a man on the moon by the end of the 1960s. The space race was on.

Among Lee’s colleagues in China’s technology community, the effect of Ke’s loss to AlphaGo was similarly galvanizing. That the program had been designed in the West was both a “challenge and an inspiration,” Lee says. Now, the AI race was on. 

In the two years since, both the Chinese government and private entrepreneurs have poured money into the field. Private investors alone sunk $4.5 billion into AI in the five-year period ending in 2017. By the end of that year, China’s venture capital investments made up 48 percent of AI venture funding globally. It was the first time China had surpassed the U.S. in total investment.

The Chinese government’s goals are explicit: By 2030, it wants China to be the center of the AI universe, leading in theory, technology and application. All the government resources pushing AI may not be invested efficiently, but the fact that everyone from China’s tech powerhouses—Huawei, Alibaba, Baidu—to individual entrepreneurs are pouring private money into the field virtually guarantees its significant growth.

FE_TrumpChina_06_993462888 A girl talks with an AI robot by Canbot during the China International Robot Show 2018 (CIROS 2018) at National Exhibition and Convention Centre in Shanghai, China. The China International Robot Show 2018 is held on July 4-7 in Shanghai. VCG/Getty

That’s true in part due to what’s happening in the AI industry globally. For decades, computer scientists at universities and in the private sector have pursued AI, but progress has been plodding. An IBM-designed computer named Deep Blue shocked the world in 1997 by defeating Garry Kasparov, then world chess champion, but the program “had few practical applications,” says Lee, who studied AI at Carnegie Mellon University in the 1980s.

AlphaGo’s victory, however, promises to be transformative. The program runs on what is called “deep learning,” which can beat humans at key, cognitive functions, including identifying faces and speech. As Lee writes in his new book, AI Superpowers, the “revolution” had looked to be about five years away. “That’s no longer true,” he says.

The key to the revolution is massive amounts of data and the computing power needed to process it. That exists: The average smartphone has more processing power than the computers that helped send Neil Armstrong to the moon in 1969, when the U.S. made good on JFK’s pledge. The data “trains” the deep-learning programs to recognize patterns, and massive computing power allows the program to sort through the data at high speeds.

Until China’s Sputnik moment, almost all the important research in the field took place in the U.S., the U.K. and Canada. The industry’s key research papers were written a full decade before Ke faced off against AlphaGo. But Lee and others believe the era of fundamental innovation in AI is now over; the future will belong to the companies and countries that can best put innovations to work.

As China demonstrated after its economic opening to the West some 40 years ago, it knows how to play catch-up in a hurry. And its massive entrepreneurial class is pragmatic and effective in learning how to build businesses out of technological innovation. Critically, in the mind of many analysts, China’s huge digital universe—where hundreds of millions of Chinese citizens spend a good chunk of their time every day—creates and captures oceans of new data about the real world, as Lee writes in AI Superpowers. Lee is not alone in believing that as AI moves into the era of “implementation,” it will be advantage China.