China’s AI Awakening中国 人工智能 的崛起
On a tropical island that marks the southern tip of China, a computer program called Lengpudashi is playing one-on-one poker against a dozen people at once, and it’s absolutely crushing them. Lengpudashi, which means “cold poker master” in Mandarin, is using a new artificial-intelligence technique to outbet and outbluff its opponents in a two-player version of Texas hold ’em.
The venue for the tournament is a modern-looking technology park in Haikou, capital of the island of Hainan. Outside, modern high-rises loom over aging neighborhoods. Those gathered to play the machine include several poker champs, some well-known Chinese investors, entrepreneurs, and CEOs, and even the odd television celebrity. The games are being broadcast online, and millions are watching. The event symbolizes a growing sense of excitement and enthusiasm for artificial intelligence in China, but there’s also a problem. Lengpudashi wasn’t made in Hainan, Beijing, or Shanghai; it was built in Pittsburgh, U.S.A.
For many in China, this simply won’t do. The country is now embarking on an unprecedented effort to master artificial intelligence. Its government is planning to pour hundreds of billions of yuan (tens of billions of dollars) into the technology in coming years, and companies are investing heavily in nurturing and developing AI talent. If this country-wide effort succeeds—and there are many signs it will—China could emerge as a leading force in AI, improving the productivity of its industries and helping it become leader in creating new businesses that leverage the technology. And if, as many believe, AI is the key to future growth, China’s prowess in the field will help fortify its position as the dominant economic power in the world.
Indeed, the country’s political and business leaders are betting that AI can jump-start its economy. In recent decades, a booming manufacturing sector—and market reforms encouraging foreign trade and investment—have helped bring hundreds of millions of people out of poverty, creating business empires and transforming Chinese society. But manufacturing growth is slowing, and the country is looking toward a future built around advanced technology (see “China Is Building a Robot Army of Model Workers”). Applying artificial intelligence may be the next step in this technology-fueled economic miracle. While many in the West fret about AI eliminating jobs and worsening wealth and income inequality, China seems to believe it can bring about precisely the opposite outcomes.
China’s AI push includes an extraordinary commitment from the government, which recently announced a sweeping vision for AI ascendancy. The plan calls for homegrown AI to match that developed in the West within three years, for China’s researchers to be making “major breakthroughs” by 2025, and for Chinese AI to be the envy of the world by 2030.
There are good reasons to believe the country can make this vision real. In the early 2000s, the government said it wanted to build a high-speed rail network that would spur technological development and improve the country’s transportation system. This train network is now the most advanced in the world. “When the Chinese government announces a plan like this, it has significant implications for the country and the economy,” says Andrew Ng, a prominent AI expert who previously oversaw AI technology and strategy at China’s biggest online search company, Baidu. “It’s a very strong signal to everyone that things will happen.”
The government’s call to action will accelerate what has already begun to happen. The country’s tech companies, led by the Internet giants Baidu, Alibaba, and Tencent, are hiring scores of AI experts, building new research centers, and investing in data centers that rival anything operated by Amazon, Google, or Microsoft. Money is also pouring into countless startups as Chinese entrepreneurs and investors spy a huge opportunity to harness AI in different industries.
China has some big advantages in AI. It has a wealth of talented engineers and scientists, for one. It also is rich in the data necessary to train AI systems. With fewer obstacles to data collection and use, China is amassing huge databases that don’t exist in other countries. The results can be seen in the growth of facial-recognition systems based on machine learning: they now identify workers at offices and customers in stores, and they authenticate users of mobile apps.
The nationwide interest in the poker tournament in Hainan reflects China’s appetite for the latest artificial-intelligence breakthroughs. Mastering even a two-player form of poker is a significant achievement for AI because, unlike many other games, poker requires players to act with limited information, and to sow uncertainty by behaving unpredictably. An optimal strategy therefore requires both careful and instinctive judgment, which are not easy qualities to give a machine. Lengpudashi impressively solved the problem by using a brilliant new game-theory algorithm, which could be very useful in many other scenarios, including financial trading and business negotiations. But Lengpudashi has received far less attention in its home country than it has in Hainan.
To explore China’s AI revolution and its implications, I’ve traveled to the heart of this boom and met with many of the key researchers, entrepreneurs, and executives. From China’s bustling capital to its factory-filled south, and from an ambitious new research center to a billion-dollar startup, one thing is clear: artificial intelligence may have been invented in the West, but you can see its future taking shape on the other side of the world.
My journey begins at MIT, one of the wellsprings of artificial intelligence. Kai-Fu Lee, a well-known Chinese AI expert and investor and one of the organizers of the Hainan tournament, has come to recruit students for a new AI institute that his company, Sinovation Ventures, is building in Beijing.
Lee gives a talk entirely in Mandarin to an auditorium packed with about 300 Chinese students. He is dressed impeccably, in an expensive-looking suit and dress shirt, and he speaks in a confident, soothing tone. The talk touches on the interwoven trends that have driven the recent rise in machine intelligence: more powerful computers, ingenious new algorithms, and huge quantities of data. He argues that China is perfectly poised to take advantage of these advances.
“The U.S. and Canada have the best AI researchers in the world, but China has hundreds of people who are good, and way more data,” he tells the audience. “AI is an area where you need to evolve the algorithm and the data together; a large amount of data makes a large amount of difference.”
In 1998 Lee founded Microsoft’s Beijing research lab, which showcased the country’s exciting talent pool (see “An Age of Ambition”). Then, in 2005, he became the founding president of Google China. Lee is now famous for mentoring young entrepreneurs, and he has more than 50 million followers on the Chinese microblogging platform Sina Weibo.
In the audience are exactly the type of prized students who would normally flock to Silicon Valley. But many are clearly taken by Lee’s message of opportunities in China. The crowd hangs on his every word, and some people clamor for autographs afterward. “Today the U.S. has a technology leadership,” Lee tells me later. “But China has a tremendous amount of potential.”
To see what this potential looks like up close, I travel to Lee’s new institute, half a world away from MIT, in Beijing’s Haidian district. The streets outside are filled with people on colorful ride-sharing bikes. I pass lots of fashionable-looking young techies as well as people delivering breakfast—ordered via smartphone, no doubt—to busy workers. At the time of my visit, a major AI event is taking place a few hundred miles to the south in Wuzhen, a picturesque town of waterways. AlphaGo, a program developed by researchers at the Alphabet subsidiary DeepMind, is playing the ancient board game Go against several top Chinese players, including the world’s number one, Ke Jie. And it’s soundly beating them.
AlphaGo’s victories in Wuzhen are followed closely in the Chinese capital. As I enter Sinovation’s institute, in fact, I notice a Go board on which engineers are testing out the moves made during some of the matches.
The location of the institute is well chosen. From the office windows, you can see the campuses of both Peking University and Tsinghua University, two of China’s top academic institutions. Sinovation provides machine-learning tools and data sets to train Chinese engineers, and it offers expertise for companies hoping to make use of AI. The institute has about 30 full-time employees so far, but the plan is to employ more than 100 by next year, and to train hundreds of AI experts each year through internships and boot camps. Right now, roughly 80 percent of the institute’s funding and projects are aimed at commercializing AI, while the rest is focused on more far-out technology research and startup incubation.
The goal isn’t to invent the next AlphaGo, though; it’s to upgrade thousands of companies across China using AI. Lee says many Chinese businesses, including the big state-owned enterprises, are technologically backward and ripe for overhaul, but they lack any AI expertise themselves. Needless to say, this presents an enormous opportunity.
Across the capital, in fact, I notice a remarkable amount of interest in artificial intelligence. In one restaurant, for instance, I find a machine that takes my picture and then supposedly uses AI to determine how healthy I am. This seems completely impossible, but the machine says I’m in great shape before suggesting that I have plenty to eat.
This fascination with the technology is reflected in Beijing’s feverish startup scene, which is already producing some formidable AI companies. One of them is SenseTime, which was founded in 2014 and is already among the world’s most valuable AI startups. Launched by researchers from the Chinese University of Hong Kong, SenseTime provides computer-vision technology to big Chinese companies, including the state-owned cellular provider China Mobile and the online retail giant JD.com. The company is now studying markets such as automotive systems. This July, SenseTime raised $410 million in funding, giving it a valuation of $1.5 billion. The entrance to SenseTime’s office features several large screens fitted with cameras. One can automatically add augmented-reality effects to a person’s face. Snapchat and Instagram offer similar gimmicks, but this one can also add effects in response to hand and body movements as well as smiles or winks.
Qing Luan, director of SenseTime’s augmented-reality group, previously developed office apps for Microsoft in Redmond, Washington. She says she returned to China because the opportunities just seemed much bigger. “We were struggling to get a thousand users; then I talked with my friend who was working at a startup in China, and she said, ‘Oh, a million users is nothing—we get that in several days,’” she recalls.
Earlier this year SenseTime’s engineers developed a novel image-processing technique for automatically removing smog and rain from photographs, and another for tracking full-body motion using a single camera. Last year they were part of a team that won a prestigious international computer-vision award.
Xiaoou Tang, SenseTime’s founder and a professor at the Chinese University of Hong Kong, is wearing a suede jacket, slacks, and glasses, and he has an intense air about him. He seems fiercely proud of his company’s achievements. Tang explains that the company’s name comes from a phonetic transcription of the name of the Shang dynasty and of its first ruler, Tang. This era, beginning around 1600 bce, was a critical age of development for the country. “China was leading the world then,” Tang says with a smile. “And in the future, we will lead again with technological innovations.”
In the United States and other Western nations, many large sectors, such as manufacturing and services, have been slow to invest in AI and change their business practices. In China, there appears to be a greater sense of urgency about adapting to the changing technology. Across just about every industry, Chinese companies are shrugging off their reputation for following Western businesses, and investing heavily in research and development. Ng, who previously led Baidu’s AI effort, says China’s business leaders understand better than most the need to embrace new trends. “The titans of industry [in China] have seen fortunes made and fortunes lost all within their lifetime,” he says. “When you see the tech trends shift, you had better move quickly, or someone else will beat you.”
Baidu anticipated the potential of artificial intelligence and sought to leverage it to reinvent its whole business. In 2014, the company created a lab dedicated to applying deep learning across its business, and in recent years, its researchers have made some significant advances. When Microsoft developed a system capable of better-than-human performance in speech recognition last year, for instance, few Western reporters realized that Baidu had done that a year earlier.
Following Baidu’s example, other Chinese tech companies are also looking to reinvent themselves with AI. The Internet leader Tencent, headquartered in the city of Shenzhen, is among them.
Shenzhen is nestled next to Hong Kong in southern China. Approaching by air, I see an armada of cargo ships moored in the South China Sea. In 1980, when Shenzhen was a small market town, it was designated China’s first Special Economic Zone, granting it unprecedented economic and regulatory freedoms. Manufacturing empires were built on the backs of migrant workers producing every imaginable product, and the population rose from 30,000 to more than 11 million. In recent years, the city has reflected China’s technological progress, and it is now home to global technology companies including the networking giant Huawei, the smartphone maker ZTE, and the electric-car company BYD.
The city’s main strip is lined with palm trees, gaudy hotels, and busy bars and restaurants. Tencent’s headquarters, in Nanshan district, spans several large buildings, and the entrance is as busy as a subway station. Stepping inside, out of stifling humidity, I begin a tour that touts Tencent’s history and achievements. And it shows that you don’t have to be first in a technology to have a big impact. In 2011, the company launched a simple messaging app, modeled on products already found in the U.S. This would evolve into WeChat, an innovative mobile platform that now supports social networking, news, games, and mobile payments. With 889 million daily active users, WeChat now has an incredible grip on China’s Internet market.
Although Tencent created an AI lab only last year, it has hired scores of researchers and opened an outpost in Seattle. The company’s researchers have already copied AI innovations from the West, including DeepMind’s AlphaGo. Tencent’s AI lab is led by Tong Zhang, a quiet man with thin glasses and a round face, who previously worked at Baidu’s AI lab and before that was a professor at Rutgers University. Zhang speaks quietly, usually after a careful pause. He explains that AI will be crucial to Tencent’s plans to grow, especially outside China. “AI is important for the next phase,” he says. “At a certain stage, you just cannot copy things. You need your own capabilities.”
I ask him if Tencent might be planning some spectacular demonstrations of AI, something like AlphaGo or Lengpudashi. Tencent owns several very popular games, including the strategy title League of Legends, which is played by more than 100 million people every month. Like Go, it requires instinctive actions, and like poker, it involves playing without a clear picture of your opponents’ standing. But it also requires planning far ahead, so it would be a worthy game for AI researchers to tackle next. “Right now, we have a bunch of small projects—some are more adventurous,” is all Zhang will say.
Tencent’s AI goals may in fact be more practical. The company has an amazing amount of conversation data thanks to WeChat and another messaging platform, called QQ. This data might be used to train machine-learning systems to hold more meaningful conversations. Making advances in language could have countless practical applications, from better document analysis and search to much smarter personal assistants. “The challenge, and also the opportunity, will be in natural language,” Zhang says.
It might be unnerving for Western nations to see a newcomer mastering an important technology, especially when the full potential of that technology remains uncertain. But it is wrong to view this story simply in terms of competition with the West.
A big problem facing both the U.S. and China is slowing economic growth. While AI may eliminate certain jobs, it also has the potential to greatly expand the economy and create wealth by making many industries far more efficient and productive. China has embraced that simple fact more eagerly and more completely than many Western nations. But there’s no reason why China’s AI-fueled economic progress should come at the expense of other countries, if those countries embrace the same technology just as keenly.
China might have unparalleled resources and enormous untapped potential, but the West has world-leading expertise and a strong research culture. Rather than worry about China’s progress, it would be wise for Western nations to focus on their existing strengths, investing heavily in research and education. The risk is missing out on an incredibly important technological shift. Yes, companies like Google and Facebook are making important strides in AI today, but this isn’t enough to reboot a whole economy. Despite the fanfare around AI, there are few economic signs—such as increased productivity—that most of the economy is taking advantage of the technology yet. Large segments of the economy beyond Silicon Valley, like medicine, service industries, and manufacturing, also need to sign on.
I can’t help thinking of the poker tournament in Hainan and reflecting that the rest of the world should take inspiration from Lengpudashi, the poker-playing AI. It’s time to follow China’s lead and go all in on artificial intelligence.
The inside story of how ChatGPT was built from the people who made it
Exclusive conversations that take us behind the scenes of a cultural phenomenon.
AI is dreaming up drugs that no one has ever seen. Now we’ve got to see if they work.
AI automation throughout the drug development pipeline is opening up the possibility of faster, cheaper pharmaceuticals.
GPT-4 is bigger and better than ChatGPT—but OpenAI won’t say why
We got a first look at the much-anticipated big new language model from OpenAI. But this time how it works is even more deeply under wraps.
The original startup behind Stable Diffusion has launched a generative AI for video
Runway’s new model, called Gen-1, can change the visual style of existing videos and movies.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.