Who will win the race for AI?

China and America are going head to to on the critical technology of the future

Dear all

Today we’re talking about the future: specifically, who will control the key technology of our time, Artificial Intelligence? America and China are locked in a desperate struggle to dominate AI, and below we’ll analyse why this is important, and who is destined to win.

In the meantime please continue to share, like, and subscribe, and leave a comment in the box if you have any input into what you’ve read. Finally, as I mentioned last Saturday, there won’t be a Chinese history/culture post this weekend, which will return on the 20th March. Thanks for reading.

****

Artificial Intelligence, or AI, is in many ways the defining technology of the future. Many claim that it will be as important as the harnessing of electricity, and something that can, in a similar way, be applied to a whole range of other industries. In other words, if AI can be dominated by any one company or nation, then it will deliver a fundamental level of control over the technologies built using it. 

Whilst much of the original work on AI was done by Americans, Britons, and Canadians, nowadays there are only two major players left: the US and China. America has led the way for long, but today China is rapidly catching up. The question is, who will become the master of AI in the future?

The last few days have shown how seriously both sides are taking the race.

In China, AI has been a focus of the recently started National People’s Congress, the country’s highest-profile annual political meeting. 5,000 delegates have come together under the watchful eye of President Xi for a two-week session to, amongst other things, make an assessment on the next national Five-Year economic plan.

The early details emerging from the session have focused heavily on science and technology, including a decree that the country will increase its annual research and development spending by more than 7% every year over the next five years. At this R&D’s core will be seven “frontier technologies”, which Beijing wants to dominate so as to keep ahead of the escalating trade war: quantum information, brain science, semiconductors, genetic research and biotechnology, clinical medicine and health, deep space/ deep sea and polar exploration – and next-generation artificial intelligence.

In Washington there have been similar, if more subdued pronouncements on the future of technology investment, in particular with regard to America’s strategic competition with China. A National Security Commission on Artificial Intelligence (NSCAI), led by former Google Chairman Eric Schmidt, has called for $32 billion in annual spending on AI so as to keep the country ahead of its strategic rival.

Kai-Fu Lee, the author of AI Superpowers and one of the world’s authorities on artificial intelligence, believes China now has the edge in AI. He notes that creating an AI superpower requires four basic building blocks: driven and ambitious entrepreneurs, well-trained AI scientists, a supportive government, and above all, masses of data. China has advantages in all four.

The fact that China has industrious entrepreneurs is staple knowledge in the tech industry. Silicon Valley has what has been termed “hustle culture”, where start-up staff are expected to work hard and forego much of their outside life to the benefit of the company. This is taken to a new level in China. Employees are expected to work “996” hours, that is 9am to 9pm six days a week, or harder if needed. Alibaba’s founder Jack Ma, one of the most successful business leaders in China, is highly supportive of the culture. Calling it a “huge blessing”, he asked in a widely publicised blog post, “How do you achieve the success you want without paying extra effort and time?”.

The early years of AI research was pushed along by some of the brightest minds ever to have worked in computing, including historical icons like the Briton Alan Turing. Nowadays, it is not the quality, but the quantity of researchers that matters, argues Kai-Fu Lee, who, as the former head of Google China, has intimate knowledge of both teams in the AI race. Strength will come from an army of trained-up engineers and entrepreneurs, and “China is training just such an army.”

Whilst the US still has the lead in the number of AI scientists and engineers (as of 2018 it was 28,500 to 18,200, in America’s favour), Chinese universities are turning out more and more graduates in AI. Just one institution, the University of Science and Technology of China, managed to grow its AI and machine learning course enrolment from 1,745 to 3,286 students between 2016 and 2018

Thanks to a supportive government, China is also able to train its algorithms in scenarios that would be unthinkable in the West, for example using AI-based facial recognition to identify jaywalkers or dispense loo-roll. Its country’s researchers are free of many of the regulations that have been blamed for slowing down AI research in the West, like Europe’s data law, GDPR.

It is however data where China has the strongest lead. 

As the saying goes, data is the new oil of the digital economy. If this is true, then China is the new Saudi Arabia, given the sheer quantity of data that it produces for its tech giants to use. 

Western firms of course have access to large amounts of online data that can be used to train AI algorithms, with Facebook claiming 190 million users in the US alone. Yet Chinese companies have an order of magnitude more. Partly this is because of the country’s huge population – it is estimated that in 2020 it had approximately 915 million internet users - but also because of the nature of the information. The Chinese companies that lead in AI have access to overwhelming amounts of offline data (i.e. data from offline sources, like shops) which makes it all the richer and more impactful in its use. The reason is the Chinese consumer. 

From about 2013, shoppers in China have embraced an astounding technological revolution, way above anything in the US or the rest of Europe. People out and about generally use their phone to pay for things rather than cash or cards. The Chinese e-commerce market is the largest in the world, with 900 million shoppers buying $4.98 trillion of online goods in 2019, and that was before the significant boost to the sector from Covid. As a World Economic Forum report noted, China is truly a market centralised around the smartphone and e-commerce. This all produces a dataset incredible in its complexity, and one ideal in its ability to train algorithms. 

None of this is to say that America is dead and buried when it comes to AI. The country has numerous advantages that are yet to fade, for example in the private sector. America has more AI start-ups than China, and in 2019 1,725 US firms in the sector received more than $1 million in funding, compared to only 224 in China. Out of the seven major companies leading the way in AI research – Amazon, Facebook, Google, Microsoft, Alibaba, Baidu, and Tencent – three are Chinese, but four are American. 

But time is running out. In last week’s newsletter we looked at how Google’s mishandling of Chinese sensitivities when it came to their national boardgame Go led to a Sputnik moment in Beijing, and the realisation that unless the country did something about AI development fast, it would be surrounded by Western technologies. It’s only now that the US has realised the enormity of what one of its company’s let loose; at least the Biden Administration is trying to do something to rectify the situation.

If AI is going to be the boon that techno-optimists claim, then a massive increase in AI R&D is likely a good thing for the majority of mankind. Neither America or China will think like that if they lose the race for AI, especially if Indermit Gill of the Brookings Institute is right in that whoever leads in artificial intelligence in 2030 will rule the world until 2100.

The good news for America is that its scientists are working on ways to reduce the importance of data as a fuel for AI development, using new training methods like “one-shot learning”, and myriad efforts to make AI that learns more like the human brain.

The bad news is that, if the China’s record in acquiring Western technologies is anything to go by, then these new techniques will be in Beijing’s hands sooner rather than later. America in turn will not take kindly to any attempts to take (by whatever means) its critical IP here. 

Given the importance of the technology to the world, there’s every chance that the race to AI supremacy is soon going to turn into a bare-knuckle fight.