China Needs to Narrow Gap With the US in AI

According to media reports, Future Today Institute founder Amy Webb released her annual tech trends report at SXSW in Austin, Texas in March 2018, which mostly focused on the impact of AI. She claimed in the report that China would "lay the groundwork to become the world's unchallenged AI hegemon" in 2018, and also mentioned that the three Chinese tech giants—Tencent, Baidu and Alibaba—deserve attention. Many may have been excited about the news, but then a question came to mind: is China really leading the world in AI technology? In an interview with CNN, Microsoft founder Bill Gates said that China may be vigorously funding AI and making strides in the field, but instead of outpacing the US, China will be in second place. His remark triggered wide discussion as to who is really the AI superpower.

When talking about development levels, the AI industry is generally divided into three areas: basic research, innovation, and application. The primary indicator for measuring the competitiveness of a company or even a nation lies in chips at the basic research level. In the current chip market, especially in the main AI chip (CPU) market, Intel holds about 71 percent of the market share, while Nvidia takes 16 percent. In other words, in the current AI chip field, US companies, with about 90 percent of the market share, have a natural monopoly advantage.

According to a report issued by the Tencent Research Institute in 2018, in terms of the number of chip makers at the basic research level, there are 14 in China and 33 in the US. Moreover, in the US, chip/processor financing totaled $31.5 billion, accounting for 31 percent and ranking first in AI financing, while in China, financing in the chip/processor field only accounted for 7.55 percent of the total AI financing. Based on the above data comparison, China's AI chips are unlikely to overtake the US in the short term.

From the perspective of human resources, of the 1.9 million global AI professionals identified by professional networking site LinkedIn, the US led the rankings with a talent pool of 850,000, while China only has 50,000 AI professionals, taking seventh place in the global AI talent rankings.

Moreover, over 70 percent of US AI professionals are working at the basic research level, while Chinese AI professionals are mostly focused on the application level in fields such as robotics, image recognition, customized marketing and autonomous driving. That is to say that Chinese AI talent is more concentrated in the application field, and the nation still lags behind the US in terms of basic chip design and development. Some media used to claim that China's AI technology had surpassed the US, citing China's number of AI publications, which have ranked first since 2014. However, considering weighted citations, the impact of Chinese AI publications only ranks 34th, far behind the fourth-place position of the US. It is obvious that China's pursuit of quantity neglects quality.

In short, although the Chinese government and relevant institutions and companies have made heavy investment in the AI industry, the chip sector at the basic research level has failed to receive enough attention in terms of the number of chip makers, investment size and talent.

It shows that the entire Chinese AI industry chain generally favors short-term profitability and focuses on technologies and applications that will only attract eyeballs from the market, shunning AI chips that require large investment and are slow to generate profits.

If Chinese companies cannot catch up at the basic chip level, it will be almost impossible for China's AI industry to overtake the US. In this sense, China should clearly recognize its gap with the US in AI technology.

This article was originally published in Global Times.