China will overtake America as leading AI superpower, warns ex-Google Beijing bigwig

Middle Kingdom just needs 'five years' to take the crown

While America today is ahead of the world in artificial intelligence, China will take the lead in five years, Google China's former president Kai-Fu Lee said today.

Lee – now founder and CEO of Sinovation Ventures, which invests in AI startups in both nations – argued the US and the Middle Kingdom each has its own strengths and weaknesses in terms of modern machine-learning development. The US has benefited from a head start, and is home to powerhouses including Apple, Facebook, and Google, and their pools of highly skilled engineering and academic talent.

However, with the help of state investment beginning a decade ago, China is now catching up, with organizations including Baidu, often compared to Google, Tencent, maker of the hugely popular WeChat app, and Alibaba, the biggest e-commerce company in the world, improving their neural networks daily.

“We often believe that the US is so far ahead, that the US invented everything,” Lee said during his keynote speech at the O’Reilly Artificial Intelligence Conference in San Francisco, USA, on Thursday.

"And it has dominated the tech scene, but a miracle happened ten years ago. That miracle is money, of course. The Chinese government has vowed to be become a world leader in AI by 2030 with its strategic roadmap dubbed ‘A Next Generation AI Development Plan’. Entrepreneurs and VCs started pouring cash into new AI startups."

A feeling of feverishness has toughened China’s tech scene. Lee said entrepreneurs fight hard to succeed, noting their grueling “9-9-6 or 9-9-7” work hours. That’s 9AM to 9PM, six or seven days per week. People are tenacious, and the competition is cutthroat, the audience was told. It’s led to a burst in "unicorn companies" - a buzzword describing rare privately owned startups that have been valued at over a billion dollars.

“There are five in China, and not that many in the US,” Lee claimed. Those five unicorns include Bitmain, a semiconductor biz focused on manufacturing custom chips to mine cryptocurrency, and Megvii, a leader in facial-recognition tech.

“Companies are no longer the X, Y, Z of China, where previously X, Y, Z were American. Now a lot of companies are copying them. China is building better products. Look at WeChat, it’s better than Facebook Messenger, and Weibo is better than Twitter.”

WeChat has morphed from a chat app to include social media sharing and mobile payments, and has been downloaded more than a billion times. Weibo is a microblogging site that provides news with more than 300 million monthly active users. China’s massive population generates more data compared to the US populace, and data is the lifeblood of modern AI: today's non-trivial neural networks need terabytes of training data to beat the competition.

At least America remains dominant in research, we're told. Chinese students strive to go to the best American universities to receive an education. Some end up staying in the US, and working for American corporations, and some decide to go back home.

Having said all this, Lee isn’t particularly a fan of portraying artificial intelligence as a clash between both countries. He wanted to “avoid Cold War metaphors,” and added: “AI isn’t like a nuclear weapon. It’s more like electricity.” He said both superpowers act as a “dual engine,” driving the progress of the technology to enable humans across the world to be more productive and useful. In theory, anyway.

Like all AI pundits, he is concerned about the displacement of jobs caused by machine-learning applications, and the enforcement of security and privacy. China is a tad obsessed with performing AI-powered facial recognition everywhere from airports to fried chicken restaurants, anywhere that has CCTV, leaving everyone constantly spied upon and tracked. However, such is life in China, where people need a decent social credit score to do anything.

“If you ask 100 people off the street in China, I’m sure they aren’t worried about it,” Lee told The Register in a separate question and answer session following his speech.

"I’m not saying that they shouldn’t be. But I think that some would think that facial recognition is good for reducing crime. Some may think that’s just the way it is, and there is nothing to do about it, and some may not even know about the potential negative consequences.

“We can’t control what companies will do. There are areas that [Sinovation Ventures] won’t touch, such as tobacco, alcohol or gambling. We try and do due diligence to invest in CEOs with professional integrity but you never know, they might pivot into other areas of interest in the future. But, yes, AI can both be used positively and negatively.” ®