Event Recap | China’s AI Ecosystem
An event organized by TAIS at Tsinghua xSpace
These are some applications which Dillon Zhou, a technology policy researcher at Tsinghua University’s Center for Industrial Development and Environmental Governance (CIDEG), discussed on Nov. 28 before delivering the keynote at an event focusing on AI policy at the university’s xSpace.
Zhou said data is key to AI, and China has the most of it. But he said quantity alone is insufficient. Big institutions and scientists need quality and diverse data to develop and commercialize applications which can harness this data.
“We have 1.4 billion people in the country. And we have so many people using smartphones. And this happens 24/7. This adds up to being a huge amount of data that gives China a tremendous advantage in AI,” Zhou said. “All firms collect data for private use. But I think it doesn’t really compare to what’s available to the government, because of what the government collects in both public and private spaces.”
Data helps AI give supercomputer the ability to recognise patterns in oceans of data. These patterns are then used to help the computer make inferences or predictions as it “learns” how to perform a task.
Chinese law only restricts free access to sensitive data on critical infrastructure, national defense, nuclear facilities, dams, megaprojects, cybersecurity and more.
But Hannah Kirk, a researcher at Peking University’s Berggruen Institute, pointed out during the panel discussion there is sometimes a cultural difference in the approach and application of AI.
“When you dig under the surface, what China says is harmonious might not be harmonious in other countries or cultures,” Kirk said.
Zhou said China differs from most countries as the government works with private companies and sometimes lets them develop the projects.
One example was Didi, Shandong University and the traffic police Jinan used AI and data last year to help ease traffic. SenseTime recently joined the government’s so-called AI “national team” to in developing facial recognition. The Chinese government started the initiative of the so-called “national team” in 2017. It also includes tech giants Baidu, Alibaba, Tencent, and Huawei as China aims to become the world leader in AI by 2030.
“It’s interesting to see that a leading company like SenseTime has a role in informing government policy,” Zhou said. “The question is whether or not their (SenseTime’s) recommendations will ultimately be binding … One wonders if the standards will be adopted.”
Zhou said China accounts for one in five tech start-ups worldwide. The U.S. has one in two. He found start-ups which get funding in China often have one or more of these qualities:
- Staff educated in the U.S. or at leading U.S. universities;
- Staff who worked on AI at a big U.S. firm; and/or
- A business model which has been proven in Western markets.
But China’s two biggest obstacles are producing computer chips and retaining AI talent. China buys nine out of ten computer chips from abroad and faces a brain drain as it struggles to keep talent graduating from top universities such as Tsinghua and Peking.
Chinese tech companies have also set up offices in Silicon Valley to entice talent to return home.
MacroPolo, a think tank part of the U.S. Paulson Institute, studied China’s growth in AI by looking at how many of its scientists the world’s most prestigious machine learning conference, NeurIPS, accepted. It found:
- China’s AI talent grew ten-fold over the last decade;
- It made up nearly one in four (24 percent) a decade later; and
- But nearly three in four (73.7 percent) now work outside China.
“So while Beijing has cultivated an army of top AI talent, well over half of that talent eventually ended up in America rather than getting hired by domestic companies and institutions,” MacroPolo said. “That’s because most of the government resources went into expanding the talent base rather than creating incentives and an environment in which they stay.”
But Andreas Pfadler, a senior AI algorithm engineer at Alibaba, said graduates are lining up the work at big Chinese tech giants such as Alibaba and Tencent.
“The competition is a bit intense, because currently basically everyone wants to do it (AI),” Pfadler said.
This article was written by Nico Gous and edited by Péter Garamvölgyi. Photos were taken by Shubh Vachher and Nico Gous.