China's edge on AI comes from its countless dialects and data

China's edge on AI comes from its countless dialects and data

AI BOT 12423 AKA 张伟 ZHANG WEI

Hi, how are you? I’m new here and I’m a robot.

Can you tell me your name please? This way, I’ll know how to address you. You can also pick a color for your chat box icon if you’d like.

China’s one billion people speak more than 200 dialects, across 23 provinces, in countless accents. Human communication across this country is challenging enough, but imagine the monumental challenge of getting its growing population of AI machines to understand every Chinese individual.

That very challenge has turned out to be a key strength in allowing China to develop what experts are calling the world’s best AI speech-recognition platform. Baidu’s Deep Speech 2 – chosen by the MIT Technology Review as one of last year’s top 10 breakthrough technologies – is emblematic of an explosion in Chinese AI and IoT innovation that will help power what UBS Equity Analyst Sundeep Gantori calls a “fourth industrial revolution.”

China’s highest state body recently announced a plan to make China the world leader in Artificial Intelligence by 2030, creating an industry worth $150 billion – and representing breathtaking investment opportunity. The nation’s research prowess is growing exponentially as brilliant Western-trained scientists return home, and billions of R&D dollars pour into the mission to achieve AI and IoT excellence. Against that backdrop, Gantori sees China enjoying a critical global advantage on its journey to AI leadership; the ability to harness data from a gigantic consumer base eager to embrace new technologies and with little hesitation to share personal information.

“AI is the engine of the rocket that takes us to the next frontier, but data is the fuel that powers the engine,” he says. “That’s something the Chinese AI-based engines have much more of than what we see in Western societies. We think that today China is in a far better position than in any of the previous industrial revolutions.” In the case of Baidu’s Deep Speech 2 platform, voice data from every corner of China are available for a technology that aims – according to Baidu speech-technology chief Liang Gao – to “change the nature of human-machine interaction.”

China’s biggest companies and regional governments are taking up the 2030 AI challenge with enthusiasm. Tencent, creator of the WeChat app, has launched an AI research lab and begun taking stakes in U.S.-based AI companies. Beyond speech recognition, Baidu is investing heavily in AI technologies to develop next-generation search and driverless vehicles; it announced it would open a new AI lab in collaboration with the government (to add to the one it already has in Silicon Valley). In Industrial IoT (IIoT), China Mobile has established a “cellular IoT open lab,” aiming to connect as many as five billion industrial devices by the end of the decade.

“Because of the new wave of investments and interest from the major Chinese IT companies, AI researchers and research activities are getting an exciting boost,” says Eric Xing, professor of machine learning at Carnegie Mellon University. Meanwhile, local governments are following state direction on AI – with the eastern city of Tianjin creating a $5 billion Artificial Intelligence fund.

According to Gantori, AI is expected to add an economic value of between $800 billion and $1.25 trillion in China by 2030. “We think 2030 is when AI technologies will become mainstream, and Chinese government projections of AI reaching a significant state is realistic around this stage,” he says. “It means that from a company or investment level, that’s the time AI will provide significant opportunities for investors in China. Unlike the previous industrial revolutions, focusing on steam, electricity and technology, in the fourth industrial revolution focusing on AI, China is only slightly behind the U.S. and is significantly closing the gap.”

The rapid strides China has made in catching up in AI leadership are underscored by UBS research showing that the number of AI-related patents filed by Chinese companies is now on a par with the U.S. Gantori cautions that many of these research filings need proof of concept. But a recent report by analytics company Elsevier and the Nikkei gave clear signs of surging quality in Chinese research, showing two institutions – the Chinese Academy of Sciences and Tsinghua University – breaking into the global top 10 of most-quoted AI-related research papers (at third and ninth respectively). Academic citations are widely seen as a gauge of quality research. The Chinese institutions were the only ones from Asia, along with Singapore’s Nanyang Technological University, in a ranking filled with the likes of Microsoft and Google.

The AI rocket taking the world to a new frontier – one of limitless technological possibility – will largely be powered by Chinese innovation. Consumer fields such as speech recognition, driverless vehicles, precision healthcare, online shopping and banking and insurance will be China’s biggest AI and IoT strengths, according to UBS research – and its technologies may permeate our daily lives sooner than we think. “By 2030, AI will be mainstream and ubiquitous,” says Gantori.

China is timing its trajectory perfectly, not only to ride but to drive the AI revolution that will transform the future: “AI gives people the feeling that China and the West are on the same starting line,” says Professor Xing. “So there is a golden chance.”

http://partners.wsj.com/ubs/invest-in-china/data-innovation-explosion/?mod=cx_picks&cx_navSource=cx_picks&cx_tag=contextual&cx_artPos=6#cxrecs_s